Pipeline v4.4.1 — Provenance-by-Construction · Validator-Enforced

R-ARF v4.4.1 Recommendations

v4.4.1 pipeline · Healthcare industry · us.anthropic.claude-sonnet-4-6 · Generated 2026-05-22 16:10:55

0 of 36
Decommission Opportunity
31 of 36
Migration Candidates
$1,432,900/yr
Run-Rate ROI
Philips Perinatal
Vendor: Philips · 10 VMs · Criticality: High · HIPAA / HITRUST
RETAIN
40%
Confidence
38%
Data Quality
Disposition: Structurally Retained On-Premises · Current Cost: $65,000/yr · Migration: Not applicable
Overview
Backlog
Provenance
Technical Details
Agents
Recommendation
RETAIN 🔒 STRUCTURAL
SRC-CAP-001 Binding cap · Ceiling: 40% · Threshold: 60%
Secondary: REPURCHASE (blocked — hardware dependency + life-safety pre-check)
Confidence
40%
Disposition is structurally certain; some documentation gaps remain to close.
ROI Summary (Python-computed, not LLM)
Structurally Retained On-Premises
Current Cost: $65,000/yr (continues unchanged)
Migration ROI: Not applicable — this system cannot move to cloud
⚙ Financials computed by provisioning_engine_v3 (Python) — not LLM-generated
Finding
Philips Perinatal is a hardware-integrated obstetric monitoring system with confirmed physical dependencies on bedside CTG fetal heart monitors — it cannot be migrated to cloud without replacing the underlying clinical hardware. The system is classified as life-safety and handles patient PHI, making it a structural retain. Priority actions are confirming the Business Associate Agreement with Philips and identifying the downstream HL7 interface consumer before any future modernization planning.
In plain terms: this app stays where it is. It is tied to physical hardware or life-safety equipment that cannot move to the cloud. The work here is compliance and lifecycle upkeep, not migration.
This is the list of work needed to move this recommendation forward. The 3 P1 items are the priority — they unlock the recommendation. P2 and P3 items improve confidence and reduce risk but are not blockers. 'Story points' measure effort, not time or cost — see the Methodology tab for how they work.
40%
Current
60%
Threshold
55%
Ceiling
34
Story Points
Why This Score
Current confidence is hard-capped at 40 by SRC-CAP-001 (phi_handled=true, baa_confirmed=UNKNOWN), which is the binding constraint — the pre-cap score of 69.0 demonstrates that underlying evidence quality is strong, but no confidence improvement is achievable until the BAA is confirmed and the cap is lifted. All backlog items target cap resolution and ROI variance reduction (currently 45%); confidence deltas are correctly set to 0 because available headroom is zero at the current ceiling.
Proceeding without BAA confirmation leaves a PHI-handling life-safety system in an unresolved HIPAA compliance posture, and without APM telemetry or architecture documentation, any future hardware refresh or SaaS migration planning rests on inferred evidence with 45% ROI variance.
Action Backlog — 6 Items · 34 Story Points · Ceiling: 55%
APP-001-001 Confirm BAA with Philips for IntelliSpace Perinatal (PHI — SRC-CAP-001 cap resolution)
P1 BAA/Legal
Retrieve or execute a signed Business Associate Agreement (BAA) with Philips covering IntelliSpace Perinatal on-premises deployment. phi_handled=true and baa_confirmed=UNKNOWN is the sole trigger for SRC-CAP-001, which hard-caps confidence at 40. Confirming the BAA resolves the cap and is the prerequisite for all subsequent confidence improvement. Engage MUSC compliance officer and Philips account
+15% confidence -0% ROI variance 👤 📋 · 8 pts
APP-001-002 Deploy 30-Day APM Telemetry Collection on All 10 Perinatal VMs
P1 Telemetry
Deploy APM agent (e.g., Dynatrace, AppDynamics, or Azure Monitor) across all 10 Philips Perinatal VMs to collect CPU, memory, disk I/O, and network time-series data over a minimum 30-day window. Current telemetry is a single RVTools snapshot (snapshot_count=1, apm_telemetry_available=false), triggering SRC-CAP-004 (ceiling 65) and SRC-RS-002 hard rule. kispnrd2 shows 82% mem_active_pct and require
+0% confidence -0% ROI variance 👤 📋 · 5 pts
APP-001-003 Retrieve Philips IntelliSpace Perinatal Contract — ACV, Renewal Date, Support Lifecycle
P1 Procurement
Procurement team must retrieve the current executed contract for Philips IntelliSpace Perinatal on-premises. Required fields: Annual Contract Value (ACV), renewal/expiration date, SLA terms, vendor support lifecycle for current version, and upgrade path to IntelliSpace Perinatal Cloud. Contract intelligence is currently MISSING, contributing a 15% ROI variance penalty. This data is also required t
+0% confidence -0% ROI variance 👤 📋 · 8 pts
APP-001-004 Identify Downstream HL7/ADT Interface Consumer from kispninterface VM
P2 Dependency Mapping
The kispninterface VM strongly implies an outbound HL7 or ADT feed, but the downstream consumer is unidentified. This creates an unmapped blast radius for any future modernization or hardware refresh. Engage clinical informatics and interface engine team to trace the HL7/ADT feed destination. Confirm whether the consumer is Epic, a clinical data repository, or another system. Update dependency cat
+0% confidence -0% ROI variance 👤 📋 · 5 pts
APP-001-005 Produce Architecture Diagram — Multi-DC Topology and CTG Hardware Integration
P2 Architecture
No architecture diagrams exist for Philips IntelliSpace Perinatal. The multi-datacenter deployment (MUSC-USB + MUSC-ART), DR failover procedures, and CTG hardware integration topology are currently INFERRED from VM naming patterns only. Engage Philips account team to obtain the ISP 8.x Installation Guide and produce a current-state architecture diagram covering: VM roles and tiers, cross-DC replic
+0% confidence -0% ROI variance 👤 📋 · 5 pts
APP-001-006 Confirm DR Requirements and Failover RTO/RPO for Life-Safety Perinatal System
P3 Architecture
DR requirements for Philips IntelliSpace Perinatal are currently INFERRED. The multi-datacenter deployment (MUSC-USB + MUSC-ART) implies geographic redundancy, but formal RTO/RPO targets, failover procedures, and clinical downtime tolerances are undocumented. Engage clinical operations and IT leadership to confirm DR requirements. Life-safety classification mandates that any DR gap be surfaced to
+0% confidence -0% ROI variance 👤 📋 · 3 pts
This tab shows the evidence behind the recommendation. The seven dimensions are scored by quality — CONFIRMED evidence counts fully, INFERRED evidence counts less. The validator table confirms each analysis step passed its checks. Green means the step is sound.
Evidence Classification — 7 Dimensions
Dimension Classification Multiplier Contribution
Telemetry SINGLE 0.50 6.00
Dependency Mapping SINGLE 0.50 5.00
Procurement Contracts MISSING 0.00 0.00
Architecture Diagrams INFERRED 0.25 1.75
Security Compliance INFERRED 0.25 1.25
Business Criticality CONFIRMED 1.00 4.00
Dr Requirements INFERRED 0.25 1.00
Applied Cap: SRC-CAP-001
Validation Reports — Per Agent
Agent Status Provenance Counts Defaults Used Rule Failures
telemetry PASS
SRC:2CUST:10COMP:4DEF:2
 None ✓ None
dependency PASS
SRC:6CUST:5COMP:7DEF:2
 None ✓ None
procurement PASS
None ✓ None
provisioning PASS_WITH_DEFAULTS
SRC:7CUST:1COMP:9DEF:1
• {'default_id': 'DEF-WL-001', 'field': 'workload_type', 'value': 'app_server', 'r
• {'default_id': 'DEF-MIG-001', 'field': 'migration_factor', 'value': 0.2, 'reason
 ✓ None
synthesizer PASS
None ✓ None
confidence_advisor PASS
COMP:16
 None ✓ None
This section contains the full technical reasoning behind the recommendation: the rules that fired, the decision path, the confidence math, and the open intelligence gaps. It is intended for analysts and engineers. Executive readers can rely on the Overview tab.
Rules Applied
These are the specific governance rules that produced this recommendation.
▸ STRUCTURAL PRE-CHECK
This app was classified RETAIN before any 6R rule was evaluated, because it has hardware integration, latency sensitivity, or a life-safety classification that physically prevents cloud migration.
▸ SRC-CAP-001 (binding)
Confidence is capped at 40 because the app handles PHI and the vendor's Business Associate Agreement is not confirmed. PHI cannot move to a cloud vendor without a confirmed BAA.
▸ SRC-CAP-002
Confidence is capped at 55 because more than half of the seven evidence dimensions are inferred or missing — the conclusion is data-thin.
▸ RULE T2 — Right-Sizing Hard Cap
Right-sizing is limited to a moderate 25% reduction because telemetry is a single snapshot. More aggressive right-sizing needs 30+ days of monitoring data.
Decision Path
The step-by-step trace of how the recommendation was reached, in the system's own terms.
PRE-CHECK EVALUATED: hardware_dependency=true (SOURCED — Philips Bedside CTG Hardware confirmed via rvtools.vms[INSPOBURGDAS].note, rvtools.vms[INSPOBURGRD2].note) AND latency_sensitive=true AND life_safety_classification=true → PRE-CHECK FIRES → strategic_recommendation=RETAIN, recommendation_status=STRUCTURAL. STOP — Rules 1-4 not evaluated. RULE 1 (REPURCHASE) BLOCKED: hardware_dependency=true is absolute blocker per pre-check refusal; saas_available=true and saas_confidence_inferred=72 are noted but cannot override structural constraint. CONFIDENCE GATE: strategic_recommendation=RETAIN → recommendation_status=STRUCTURAL → confidence_gate_applied=false. displayed_primary=RETAIN (structural, not gate-forced). CONFIDENCE SCORING: base=50 + telemetry(SINGLE,12,0.50)=6.0 + dependency(SINGLE,10,0.50)=5.0 + procurement(MISSING,8,0.00)=0.0 + architecture(INFERRED,7,0.25)=1.75 + security(INFERRED,5,0.25)=1.25 + business_criticality(CONFIRMED,4,1.00)=4.0 + dr(INFERRED,4,0.25)=1.0 → pre_cap=69.0. CAP EVALUATION: SRC-CAP-001 fires (phi_handled=true, baa_confirmed=UNKNOWN≠true) → cap=40 BINDING. SRC-CAP-002 fires (4/7 dimensions INFERRED/MISSING) → cap=55. SRC-CAP-004 fires (snapshot_count=1, apm=false) → cap=65. Lowest wins: SRC-CAP-001 → final_confidence=40.
Cap Evaluation (all fired caps): SRC-CAP-001: PHI handled + BAA not confirmed → ceiling 40 (ceiling=40) ← BINDING SRC-CAP-002: ≥4 of 7 dimensions INFERRED/MISSING → ceiling 55 (ceiling=55)
Confidence Scoring Breakdown
How the confidence percentage was calculated, dimension by dimension.
Base score: 50 Evidence contributions: Telemetry SINGLE +6.00 Dependency Mapping SINGLE +5.00 Procurement Contracts MISSING +0.00 Architecture Diagrams INFERRED +1.75 Security Compliance INFERRED +1.25 Business Criticality CONFIRMED +4.00 Dr Requirements INFERRED +1.00 Pre-cap score: 69.0 Binding cap applied: SRC-CAP-001 Final confidence: 40%
Intelligence Gaps
Specific things we don't yet know. Each gap is a reason the confidence score is held back.
BAA Status with Philips — Unconfirmed
phi_handled=true but baa_confirmed=UNKNOWN. HIPAA requires a signed Business Associate Agreement with Philips for any PHI processing. This is a compliance gap independent of disposition and must be resolved immediately. Triggers SRC-CAP-001, capping confidence at 40.
HL7/ADT Interface Consumer Identity — Unconfirmed
kispninterface VM strongly implies an outbound HL7 or ADT feed, but the downstream consumer is unidentified. This creates an unmapped blast radius. Any future modernization or hardware refresh planning requires this dependency to be confirmed.
Contract Intelligence — Unavailable
No contract data available for Philips IntelliSpace Perinatal on-premises. Renewal dates, SLA terms, vendor support lifecycle, and upgrade path to cloud version are unknown. Procurement team must retrieve current contract.
Architecture Documentation — Absent
No architecture diagrams available. Multi-datacenter deployment topology (MUSC-USB + MUSC-ART), DR failover procedures, and CTG hardware integration architecture are inferred from VM naming patterns only. Required for any future hardware refresh planning.
APM Telemetry — Unavailable
No APM data available. Right-sizing analysis is limited to single RVTools memory snapshot. kispnrd2 shows 82% mem_active_pct — the only VM approaching pressure thresholds — and warrants investigation to determine if this reflects sustained load or a snapshot artifact.
Technical Detail
A fuller technical explanation of the recommendation and its constraints.
The dependency agent confirmed hardware_dependency=true, latency_sensitive=true, and life_safety_classification=true via SOURCED evidence from RVTools VM notes identifying Philips Bedside Fetal Heart Monitors (CTG Hardware) as upstream physical dependencies. The v4.1 structural pre-check fires unconditionally on this combination, overriding the procurement agent's finding of saas_available=true (Philips IntelliSpace Perinatal Cloud, saas_confidence_inferred=72). The telemetry agent's snapshot_count=1 and apm_telemetry_available=false triggered SRC-RS-002 right-sizing hard rule, and the procurement agent returned baa_confirmed=UNKNOWN, triggering SRC-CAP-001 to cap confidence at 40. REPURCHASE is explicitly refused per v4.1 pre-check refusal rules for hardware-integrated clinical devices.
Secondary Path
The fallback option if the primary recommendation cannot proceed.
Secondary Path — REPURCHASE (blocked — hardware dependency + life-safety pre-check)
Philips IntelliSpace Perinatal Cloud is a viable long-term modernization target, but only after a clinical hardware refresh program replaces the bedside CTG monitors with cloud-compatible equivalents. This should be tracked as a future-state initiative in the portfolio roadmap, not an active migration.
Right-Sizing Notes
Why the cloud sizing estimate is set where it is, based on the telemetry available.
Fleet avg mem_active_pct = 24.4% across 10 VMs (all powered on, all Windows Server 2019, no memory pressure). Under Rule T3 alone, 24.4% < 25% would classify as OVERSIZED (SRC-RS-003, 0.40 reduction). However, Rule T2 fires unconditionally because snapshot_count=1 and apm_telemetry_available=false; OVERSIZED and ZOMBIE tiers are forbidden. Tier capped at MODERATE (SRC-RS-002, 0.25 reduction). CPU utilization data is null for all 10 VMs; no CPU-based sizing analysis is possible. Notable outlier: kispnrd2 (MUSC-ART) shows 82.0% mem_active_pct — this VM appears well-utilized or potentially undersized and must be excluded from any right-sizing action pending further telemetry. Application is a clinical fetal heart monitoring system (Philips IntelliSpace Perinatal); life_safety_classification is UNKNOWN (catalog field null — TRI-STATE rule applied, not inferred). Any right-sizing action must be validated with Philips and clinical/biomedical stakeholders before implementation. Recommend 30+ day APM collection to enable definitive tier assignment.
This tab shows the raw output of each analysis agent — the specialized components that examine telemetry, dependencies, contracts, and cost. It is the most detailed and most technical view. Most readers will find the Overview and Backlog tabs sufficient.
Telemetry Agent
Utilization Profilemoderate
Snapshot Count1
APM AvailableFalse
Hard Rule TriggeredTrue
Right-Sizing Opp.True
Agent Confidence48%
Telemetry — VM Counts
Powered Off VMs0
Flagged VMs0
Mem Active % (avg)24.4%
Data WindowNone
Inventory Source
Dependency Agent
Epic IntegrationFalse
PHI HandledTrue
Hardware DependencyTrue
Latency SensitiveTrue
Blast Radiuscritical
Migration RiskHIGH
Conf (Inferred)82%
Conf (Confirmed)61%
Verificationsingle_source
Cloud Connectivity
ComplianceHIPAA / HITRUST
Upstream (2)
Philips Bedside Fetal Heart Monitors (CTG Hardware): confirmed: rvtools.vm_notes — INSPOBURGDAS, INSPOBURGRD2 ('Philips Fetal Heart Monitor Orangeburg')
Hospital Network Infrastructure (LAN/VLAN — MUSC-USB, MUSC-ART): confirmed: rvtools.datacenters — ['MUSC-ART', 'MUSC-USB']; multi-datacenter deployment implies cross-DC replication…
Downstream (2)
Unknown HL7/ADT Interface Consumer (inferred from kispninterface VM): inferred: rvtools.vm_notes — kispninterface VM name pattern 'interface' strongly implies HL7 or ADT outbound feed to…
Epic EMR (candidate — not confirmed): inferred: industry pattern — Philips ISP commonly integrates with Epic via HL7 ADT/ORU feeds in academic medical center…
Procurement Agent
SaaS Available—
SaaS VendorPhilips
SaaS ProductPhilips IntelliSpace Perinatal Cloud
Repurchase ViableINSUFFICIENT_EVIDENCE
BAA ConfirmedUNKNOWN
SaaS Confidence
Inferred Score72%
Confirmed Score0%
Verificationsingle_source
AWS Equiv.None
Provisioning Agent (Python engine — not LLM)
Current VMs10
Recommended VMs10
Right-Sizing0% reduction
Current TCO/yr$65,000
Projected Cloud TCO$52,000
Migration One-Time$10,400
Dual-Running Cost$39,000
N/A — RETAIN
Year-1 Net
N/A — RETAIN
Steady-State
N/A — RETAIN
3-Year
Migration cost figures are not computed for structurally retained applications.
AWS TargetAmazon EC2
Azure TargetAzure Virtual Machines
Cost Modeldefault_multipliers
DEF: {'default_id': 'DEF-WL-001', 'field': 'workload_type', 'value': 'app_server', 'reason': 'workload_type indeterminable'}
DEF: {'default_id': 'DEF-MIG-001', 'field': 'migration_factor', 'value': 0.2, 'reason': 'migration_complexity not provided'}
DEF: {'default_id': 'DEF-COST-001', 'field': 'cost_model_basis', 'value': 'default_multipliers', 'reason': 'no customer cloud quote provided; using SRC-TCO-* and SRC-MULT-* registry defaults'}
Philips Patient Monitoring
Vendor: Philips · 20 VMs · Criticality: High · HIPAA / HITRUST — PHI handling status un
RETAIN
55%
Confidence
25%
Data Quality
Disposition: Structurally Retained On-Premises · Current Cost: $130,000/yr · Migration: Not applicable
Overview
Backlog
Provenance
Technical Details
Agents
Recommendation
RETAIN 🔒 STRUCTURAL
SRC-CAP-002 Binding cap · Ceiling: 55% · Threshold: 60%
Secondary: REPURCHASE (conditional — pending hardware dependency resolution and BAA confirmation)
Confidence
55%
Disposition is structurally certain; some documentation gaps remain to close.
ROI Summary (Python-computed, not LLM)
Structurally Retained On-Premises
Current Cost: $130,000/yr (continues unchanged)
Migration ROI: Not applicable — this system cannot move to cloud
⚙ Financials computed by provisioning_engine_v3 (Python) — not LLM-generated
Finding
Philips Patient Monitoring is a hardware-integrated, life-safety clinical system with direct dependencies on Philips IntelliVue and GE bedside monitoring hardware units — it cannot be migrated or replaced without a formal clinical device transition program. The system is structurally retained on-premises until a validated hardware-inclusive migration pathway is established. A secondary REPURCHASE pathway via Philips HealthSuite exists but requires hardware dependency resolution, BAA confirmation, and clinical validation before it can be actioned.
In plain terms: this app stays where it is. It is tied to physical hardware or life-safety equipment that cannot move to the cloud. The work here is compliance and lifecycle upkeep, not migration.
This is the list of work needed to move this recommendation forward. The 3 P1 items are the priority — they unlock the recommendation. P2 and P3 items improve confidence and reduce risk but are not blockers. 'Story points' measure effort, not time or cost — see the Methodology tab for how they work.
55%
Current
60%
Threshold
63%
Ceiling
29
Story Points
Why This Score
The synthesizer score of 55 is bound by SRC-CAP-002 (majority of dimensions INFERRED or MISSING: dependency_mapping=INFERRED, procurement_contracts=MISSING, architecture_diagrams=MISSING, security_compliance=INFERRED, dr_requirements=INFERRED — 5 of 7 dimensions), which imposes a hard ceiling of 55 regardless of the pre-cap calculated score of 62.75. Because current_confidence equals the ceiling, no backlog item can produce a confidence climb in this run; all items target cap resolution, variance reduction, and compliance remediation that will unlock headroom in a future synthesizer re-run once the binding cap is lifted.
The structural RETAIN classification is well-supported by life-safety and hardware-dependency signals, but the active HIPAA compliance exposure (unconfirmed BAA, two EOL Windows Server 2012 VMs on a PHI-handling system) represents an immediate regulatory risk that must be remediated independent of any migration decision.
Action Backlog — 6 Items · 29 Story Points · Ceiling: 63%
APP-002-001 Confirm or execute Philips BAA — resolve PHI/BAA compliance gap
P1 BAA/Legal
Locate any existing BAA with Philips covering the on-premises Patient Monitoring deployment and the HealthSuite SaaS platform. If no BAA exists, engage Philips legal/compliance team to execute one. Confirm canonical phi_handled=true in the application catalog by updating the field from the non-standard 'phi_handling' key. Obtain a signed BAA artifact and cite it in the catalog record. This resolve
+0% confidence -0% ROI variance 👤 📋 · 3 pts
APP-002-002 Confirm hardware dependency scope via clinical engineering and vendor documentation
P1 Dependency Mapping
Obtain formal confirmation of the hardware dependency scope from clinical engineering and Philips vendor documentation. Specifically: (1) Confirm whether Philips IntelliVue bedside units and GE bedside hardware units require direct on-premises server connectivity or can operate via a cloud gateway. (2) Obtain network diagrams or integration specifications for the physio-server tier (patmonphysio01
+0% confidence -0% ROI variance 👤 📋 · 5 pts
APP-002-003 Obtain Philips contract terms — ACV, renewal date, and SLA commitments
P1 Procurement
Retrieve the active Philips support and maintenance contract for the on-premises Patient Monitoring deployment. Capture: (1) Annual contract value (ACV), (2) Renewal date, (3) SLA terms and uptime commitments, (4) Whether the contract includes any HealthSuite migration incentives or upgrade pathways. This resolves the procurement_contracts=MISSING dimension, reduces ROI variance by 15 percentage p
+8% confidence -0% ROI variance 👤 📋 · 3 pts
APP-002-004 Produce formal architecture diagram for the 20-VM patient monitoring topology
P2 Architecture
Engage the infrastructure team and Philips POC (Matthew Corson) to produce a formal architecture diagram covering: (1) VM role assignments for all 20 nodes (physio servers, web tier, IBE, DB cluster, patmonbase powered-off VM), (2) Network topology and VLAN segmentation, (3) Data flows between bedside hardware, physio servers, web tier, and DB cluster, (4) DR/HA configuration for the dual-node DB
+8% confidence -0% ROI variance 👤 📋 · 5 pts
APP-002-005 Remediate Windows Server 2012 EOL on IBE VMs — active HIPAA security exposure
P2 Security/Compliance
patmon-ibe-tst and patmonIBEapp are running Windows Server 2012 (64-bit), which reached end of extended support on October 10, 2023. On a PHI-handling system, unpatched OS constitutes an active HIPAA Security Rule violation (45 CFR §164.312). Required actions: (1) Upgrade both VMs to Windows Server 2019 or 2022, OR (2) Apply Microsoft Extended Security Updates (ESU) as a compensating control while
+8% confidence -0% ROI variance 👤 📋 · 8 pts
APP-002-006 Deploy 30-day APM telemetry collection to resolve single-snapshot right-sizing constraint
P3 Telemetry
Deploy APM tooling (e.g., Dynatrace, AppDynamics, or Azure Monitor) across the 20-VM fleet to collect a minimum 30-day time-series of CPU utilization, memory active percentage, and network I/O. Current telemetry is a single RVTools snapshot with no CPU data (cpu_avg_pct null on all 20 VMs). This prevents right-sizing beyond MODERATE tier (SRC-RS-002) and keeps SRC-CAP-004 active. Note: life_safety
+0% confidence -0% ROI variance 👤 📋 · 5 pts
This tab shows the evidence behind the recommendation. The seven dimensions are scored by quality — CONFIRMED evidence counts fully, INFERRED evidence counts less. The validator table confirms each analysis step passed its checks. Green means the step is sound.
Evidence Classification — 7 Dimensions
Dimension Classification Multiplier Contribution
Telemetry SINGLE 0.50 6.00
Dependency Mapping INFERRED 0.25 2.50
Procurement Contracts MISSING 0.00 0.00
Architecture Diagrams MISSING 0.00 0.00
Security Compliance INFERRED 0.25 1.25
Business Criticality SINGLE 0.50 2.00
Dr Requirements INFERRED 0.25 1.00
Applied Cap: SRC-CAP-002
Validation Reports — Per Agent
Agent Status Provenance Counts Defaults Used Rule Failures
telemetry PASS
SRC:2CUST:9COMP:5DEF:1
 None ✓ None
dependency PASS
SRC:12CUST:1COMP:8DEF:3
 None ✓ None
procurement PASS
None ✓ None
provisioning PASS_WITH_DEFAULTS
SRC:7CUST:1COMP:9DEF:1
• {'default_id': 'DEF-WL-001', 'field': 'workload_type', 'value': 'app_server', 'r
• {'default_id': 'DEF-MIG-001', 'field': 'migration_factor', 'value': 0.2, 'reason
 ✓ None
synthesizer PASS
None ✓ None
confidence_advisor PASS
COMP:15
 None ✓ None
This section contains the full technical reasoning behind the recommendation: the rules that fired, the decision path, the confidence math, and the open intelligence gaps. It is intended for analysts and engineers. Executive readers can rely on the Overview tab.
Rules Applied
These are the specific governance rules that produced this recommendation.
▸ STRUCTURAL PRE-CHECK
This app was classified RETAIN before any 6R rule was evaluated, because it has hardware integration, latency sensitivity, or a life-safety classification that physically prevents cloud migration.
▸ SRC-CAP-002 (binding)
Confidence is capped at 55 because more than half of the seven evidence dimensions are inferred or missing — the conclusion is data-thin.
▸ SRC-CAP-004
Confidence is capped at 65 because telemetry is a single point-in-time snapshot with no ongoing monitoring — right-sizing needs usage patterns over time.
▸ RULE T2 — Right-Sizing Hard Cap
Right-sizing is limited to a moderate 25% reduction because telemetry is a single snapshot. More aggressive right-sizing needs 30+ days of monitoring data.
Decision Path
The step-by-step trace of how the recommendation was reached, in the system's own terms.
PRE-CHECK evaluated first: hardware_dependency=true (SOURCED: dependency agent, inferred from vendor architecture + RVTools VM names) AND latency_sensitive=true (SOURCED: dependency agent) AND life_safety_classification=true (SOURCED: dependency agent) → PRE-CHECK FIRES → strategic_recommendation=RETAIN, recommendation_status=STRUCTURAL. Rule 1 (REPURCHASE) explicitly refused: hardware_dependency=true AND life_safety_classification=true — Philips HealthSuite SaaS availability does not override structural constraint per v4.0 pre-check refusal clause. Rules 2–5 not evaluated. Confidence scoring: base=50 + telemetry(SINGLE,12×0.50=6.0) + dependency(INFERRED,10×0.25=2.5) + procurement(MISSING,8×0.00=0.0) + architecture(MISSING,7×0.00=0.0) + security(INFERRED,5×0.25=1.25) + criticality(SINGLE,4×0.50=2.0) + dr(INFERRED,4×0.25=1.0) = 62.75 pre-cap. SRC-CAP-002 fires (5 of 7 dimensions INFERRED/MISSING → majority) → cap 55. SRC-CAP-004 fires (snapshot_count=1, apm_telemetry_available=false) → cap 65. Lowest cap = SRC-CAP-002 at 55. final_confidence=55. Gate evaluation: recommendation_status=STRUCTURAL (RETAIN), confidence_gate_applied=false. ROI variance: base(5) + procurement_penalty(MISSING,+15) + arch_penalty(MISSING,+15) + dep_penalty(INFERRED,+10) + dr_penalty(INFERRED,+5) = 50% (telemetry=SINGLE, no penalty; under 60% cap).
Cap Evaluation (all fired caps): SRC-CAP-002: ≥4 of 7 dimensions INFERRED/MISSING → ceiling 55 (ceiling=55) ← BINDING SRC-CAP-004: Single snapshot, no APM → ceiling 65 (ceiling=65)
Confidence Scoring Breakdown
How the confidence percentage was calculated, dimension by dimension.
Base score: 50 Evidence contributions: Telemetry SINGLE +6.00 Dependency Mapping INFERRED +2.50 Procurement Contracts MISSING +0.00 Architecture Diagrams MISSING +0.00 Security Compliance INFERRED +1.25 Business Criticality SINGLE +2.00 Dr Requirements INFERRED +1.00 Pre-cap score: 62.75 Binding cap applied: SRC-CAP-002 Final confidence: 55%
Intelligence Gaps
Specific things we don't yet know. Each gap is a reason the confidence score is held back.
BAA confirmation absent — PHI handling status unconfirmed
phi_handled=UNKNOWN and baa_confirmed=UNKNOWN for a system processing real-time patient vitals. HIPAA BAA with Philips must be confirmed or executed. Blocks any future REPURCHASE evaluation and creates current compliance exposure.
Hardware dependency scope unconfirmed — bedside unit integration inferred only
Philips IntelliVue and GE bedside unit dependencies are inferred from vendor architecture patterns and VM naming conventions, not confirmed via network diagrams, clinical engineering documentation, or vendor integration specifications. Structural RETAIN classification depends on this dependency being real — confirmation required.
Epic EHR alarm/event feed dependency unresolved
catalog.epic_integration=false contradicts healthcare industry pattern (Philips PIIC iX supports HL7 ADT/alarm feeds to Epic). If an undocumented Epic integration exists, it creates an additional migration constraint. Requires clinical informatics team confirmation.
Architecture diagrams absent — 20-VM topology inferred from naming conventions
No formal architecture documentation available. VM roles (physio servers, web tier, IBE, DB cluster) inferred from naming patterns. Without confirmed topology, blast radius and migration sequencing cannot be validated.
Contract intelligence insufficient — no active contract terms available
No contract renewal dates, SLA terms, or vendor support commitments available. Cannot assess whether current Philips support contract covers the 20-VM on-prem deployment or whether HealthSuite migration is contractually incentivized.
Windows Server 2012 EOL on IBE VMs — active security exposure
patmon-ibe-tst and patmonIBEapp are running Windows Server 2012 (64-bit), which reached end of extended support in October 2023. These VMs are no longer receiving security patches. Immediate remediation required — upgrade to Windows Server 2019/2022 or isolate from network.
patmonbase VM powered off — purpose and decommission status unknown
1 of 20 VMs (patmonbase, 4 vCPU / 12 GiB RAM, Windows Server 2016) is powered off with no note or flag. Unknown whether this is a decommissioned standby, a failed node, or a planned spare. Requires POC (Matthew Corson) confirmation.
Technical Detail
A fuller technical explanation of the recommendation and its constraints.
The pre-check fires unconditionally: hardware_dependency=true, latency_sensitive=true, and life_safety_classification=true are all confirmed by the dependency agent. Philips Patient Monitoring aggregates real-time physiological data from Philips IntelliVue and GE bedside hardware units — the software layer cannot be decoupled from the physical monitoring hardware without a formal clinical device replacement program. The procurement agent identifies Philips HealthSuite as a SaaS alternative (saas_confidence_inferred=68), but Rule 1 (REPURCHASE) is explicitly refused for hardware-integrated life-safety systems per v4.0 governance. The system is structurally retained at MUSC-USB until a hardware-inclusive migration pathway is validated through clinical engineering and vendor engagement.
Secondary Path
The fallback option if the primary recommendation cannot proceed.
Secondary Path — REPURCHASE (conditional — pending hardware dependency resolution and BAA confirmation)
REPURCHASE via Philips HealthSuite is the logical long-term pathway if MUSC undertakes a clinical device refresh program that replaces physical bedside units with HealthSuite-compatible hardware. This would require: (1) hardware dependency resolution confirmed by clinical engineering, (2) BAA execution with Philips for HealthSuite, (3) clinical validation and FDA/ONC compliance review, and (4) Epic HL7 alarm feed dependency resolution.
Right-Sizing Notes
Why the cloud sizing estimate is set where it is, based on the telemetry available.
Fleet avg mem_active_pct is 12.1%, which under Rule T3 would qualify as OVERSIZED (SRC-RS-003, <25% threshold). However, snapshot_count=1 AND apm_telemetry_available=false trigger Rule T2 hard cap. OVERSIZED (SRC-RS-003) and ZOMBIE (SRC-RS-004) tiers are FORBIDDEN. Tier capped at MODERATE (SRC-RS-002, 0.25 reduction). Single-snapshot data cannot distinguish genuinely idle VMs from those sampled between burst cycles. CPU utilization data is entirely absent (cpu_avg_pct null on all 20 VMs). Two VMs (patmon-ibe-tst, patmonIBEapp) run Windows Server 2012, which reached EOL October 2023; phi_handling=true makes these active HIPAA Security Rule violations requiring immediate remediation.
This tab shows the raw output of each analysis agent — the specialized components that examine telemetry, dependencies, contracts, and cost. It is the most detailed and most technical view. Most readers will find the Overview and Backlog tabs sufficient.
Telemetry Agent
Utilization Profilemoderate
Snapshot Count1
APM AvailableFalse
Hard Rule TriggeredTrue
Right-Sizing Opp.True
Agent Confidence45%
Telemetry — VM Counts
Powered Off VMs1
Flagged VMs0
Mem Active % (avg)12.1%
Data Window1 days
Inventory Source
Dependency Agent
Epic IntegrationFalse
PHI HandledUNKNOWN
Hardware DependencyTrue
Latency SensitiveTrue
Blast Radiuscritical
Migration RiskHIGH
Conf (Inferred)72%
Conf (Confirmed)38%
Verificationsingle_source
Cloud Connectivity
ComplianceHIPAA / HITRUST
Upstream (4)
Philips Bedside Monitors / IntelliVue Hardware Units: inferred: vendor-architecture-pattern (Philips PIIC iX requires physical bedside units as upstream data sources)
GE Bedside Hardware (partial integration): inferred: rvtools.vm_notes.patmonphysio01d='Patmon GE', rvtools.vm_notes.patmonphysio02d='Patmon GE',…
IBE (Integration Bus Engine) — patmon-ibe-tst / patmonIBEapp: confirmed: rvtools.vm_notes.patmon-ibe-tst='Patcom IBE test database server. POC Desiree Tillman'; VM patmonIBEapp…
Active Directory / Windows Domain Services: inferred: Windows Server OS across all 20 VMs; Client/Server application_type implies domain authentication dependency
Downstream (3)
Clinical Staff Workstations / Central Monitoring Stations: inferred: vendor-architecture-pattern (PIIC iX central station clients consume physio-server data streams)
Epic EHR (alarm/event feed — potential): inferred: healthcare industry pattern; Philips PIIC iX supports HL7 ADT/alarm feeds to Epic. catalog.epic_integration=fa…
Patmon Prime Database Tier (patmonprimedb-c / patmonprimedb-d): confirmed: rvtools.vm_names[patmonprimedb-c, patmonprimedb-d]; dual-node DB cluster (c=current, d=dev/DR) confirmed by…
Procurement Agent
SaaS Available—
SaaS VendorPhilips
SaaS ProductPhilips HealthSuite
Repurchase ViableINSUFFICIENT_EVIDENCE
BAA ConfirmedUNKNOWN
SaaS Confidence
Inferred Score68%
Confirmed Score0%
Verificationsingle_source
AWS Equiv.None
Provisioning Agent (Python engine — not LLM)
Current VMs20
Recommended VMs20
Right-Sizing0% reduction
Current TCO/yr$130,000
Projected Cloud TCO$104,000
Migration One-Time$20,800
Dual-Running Cost$78,000
N/A — RETAIN
Year-1 Net
N/A — RETAIN
Steady-State
N/A — RETAIN
3-Year
Migration cost figures are not computed for structurally retained applications.
AWS TargetAmazon EC2
Azure TargetAzure Virtual Machines
Cost Modeldefault_multipliers
DEF: {'default_id': 'DEF-WL-001', 'field': 'workload_type', 'value': 'app_server', 'reason': 'workload_type indeterminable'}
DEF: {'default_id': 'DEF-MIG-001', 'field': 'migration_factor', 'value': 0.2, 'reason': 'migration_complexity not provided'}
DEF: {'default_id': 'DEF-COST-001', 'field': 'cost_model_basis', 'value': 'default_multipliers', 'reason': 'no customer cloud quote provided; using SRC-TCO-* and SRC-MULT-* registry defaults'}
PaceArt
Vendor: Biotronik · 5 VMs · Criticality: High · HIPAA / HITRUST — PHI-bearing life-safet
RETAIN
40%
Confidence
22%
Data Quality
Disposition: Structurally Retained On-Premises · Current Cost: $32,500/yr · Migration: Not applicable
Overview
Backlog
Provenance
Technical Details
Agents
Recommendation
RETAIN 🔒 STRUCTURAL
SRC-CAP-001 Binding cap · Ceiling: 40% · Threshold: 60%
Secondary: REHOST (conditional — requires hardware_dependency=false confirmed by Biotronik)
Confidence
40%
Disposition is structurally certain; some documentation gaps remain to close.
ROI Summary (Python-computed, not LLM)
Structurally Retained On-Premises
Current Cost: $32,500/yr (continues unchanged)
Migration ROI: Not applicable — this system cannot move to cloud
⚙ Financials computed by provisioning_engine_v3 (Python) — not LLM-generated
Finding
PaceArt is a Biotronik cardiac device management system classified as life-safety, handling PHI for patients with active pacemakers and ICDs. No SaaS alternative exists and Biotronik cloud-hosted deployment compatibility is unconfirmed, making migration inadvisable without vendor engagement. The system is retained on-premises, but carries critical compliance debt: all five VMs run end-of-life Windows Server 2012 R2, the production database runs end-of-life SQL Server 2008, and BAA status is unconfirmed — each of these requires immediate remediation regardless of cloud disposition.
In plain terms: this app stays where it is. It is tied to physical hardware or life-safety equipment that cannot move to the cloud. The work here is compliance and lifecycle upkeep, not migration.
This is the list of work needed to move this recommendation forward. The 2 P1 items are the priority — they unlock the recommendation. P2 and P3 items improve confidence and reduce risk but are not blockers. 'Story points' measure effort, not time or cost — see the Methodology tab for how they work.
40%
Current
60%
Threshold
55%
Ceiling
32
Story Points
Why This Score
PaceArt is capped at 40 by SRC-CAP-001 because phi_handled=true and baa_confirmed=UNKNOWN — the catalog baa_valid field is null, which mandates UNKNOWN under v3.1 tri-state rules, and this condition alone imposes a hard ceiling of 40 regardless of other evidence quality. The pre-cap score of 61.75 reflects that the application is well-understood in terms of its clinical domain and life-safety classification, but five of seven evidence dimensions are INFERRED or MISSING, compounding the BAA gap with SRC-CAP-002 and SRC-CAP-003 as sequential caps that will bind once SRC-CAP-001 is resolved.
Proceeding without BAA confirmation for a PHI-bearing life-safety cardiac device management system exposes the organization to active HIPAA enforcement risk independent of any cloud migration decision, and the unresolved hardware_dependency leaves the structural disposition of PaceArt formally incomplete.
Action Backlog — 6 Items · 32 Story Points · Ceiling: 55%
APP-003-001 Execute BAA with Biotronik and confirm PHI handling agreement
P1 BAA/Legal
Obtain and execute a signed Business Associate Agreement (BAA) with Biotronik covering PaceArt's handling of PHI in the on-premises deployment. Catalog field baa_valid is null, triggering SRC-CAP-001 and capping confidence at 40. Resolving baa_confirmed=true is the single action that lifts the binding cap. Deliverable: executed BAA document filed in the compliance artifact repository with date and
+15% confidence -0% ROI variance 👤 📋 · 8 pts
APP-003-002 Obtain Biotronik written confirmation of hardware dependency and cloud deployment support
P1 Vendor Engagement
Submit a formal technical inquiry to Biotronik requesting: (1) whether PaceArt Optima requires co-located physical hardware programmers (e.g., ICS 3000, RENAMIC) connected to the application VMs, (2) whether cloud-hosted or virtualized deployment is supported under Biotronik's product warranty and FDA regulatory clearance, and (3) supported OS and database versions for current PaceArt Optima relea
+0% confidence -0% ROI variance 👤 📋 · 8 pts
APP-003-003 Conduct full interface and dependency mapping including HL7 consumers
P2 Dependency Mapping
Perform a structured dependency mapping exercise to confirm all upstream and downstream integrations for PaceArt. Current dependency_mapping classification is INFERRED. Specific gaps: (1) HL7 interface inventory — epic_integration=false per catalog but PaceArt Optima supports HL7 feeds to scheduling, billing, and EMR systems that are not inventoried; (2) confirm whether MUSC-ART datacenter hosts a
+0% confidence -0% ROI variance 👤 📋 · 5 pts
APP-003-004 Retrieve and review Biotronik PaceArt procurement contract and licensing terms
P2 Procurement
Locate and review the executed Biotronik PaceArt contract. contract_intelligence=INSUFFICIENT_EVIDENCE; contract_acv and contract_renewal_date are both null. Required outputs: (1) ACV and total contract value, (2) renewal date and notice period, (3) licensing terms for virtualized or cloud-hosted deployment — Biotronik may restrict deployment to bare-metal or specific hypervisor configurations, (4
+0% confidence -0% ROI variance 👤 📋 · 3 pts
APP-003-005 Document current architecture and EOL remediation plan for PaceArt infrastructure
P2 Architecture
Produce a current-state architecture diagram for the PaceArt 5-VM deployment and a formal EOL remediation plan. architecture_diagrams=MISSING. The EOL stack is critical: all 5 VMs run Windows Server 2012 R2 (EOL Oct 2023) and the production DB runs SQL Server 2008 (ESUs expired July 2022) — both are active HIPAA Security Rule risks in a PHI-bearing life-safety system. The architecture document mus
+0% confidence -0% ROI variance 👤 📋 · 5 pts
APP-003-006 Document DR requirements and validate contingency posture for life-safety cardiac system
P3 Security/Compliance
Establish formal DR requirements for PaceArt including RTO, RPO, and contingency procedures. dr_requirements=MISSING for a life-safety system is a HIPAA contingency planning gap (45 CFR §164.308(a)(7)). The dual-datacenter footprint (MUSC-ART production, MUSC-USB test) suggests some HA awareness but no formal DR posture is confirmed. Required outputs: (1) documented RTO/RPO targets approved by cli
+0% confidence -0% ROI variance 👤 📋 · 3 pts
This tab shows the evidence behind the recommendation. The seven dimensions are scored by quality — CONFIRMED evidence counts fully, INFERRED evidence counts less. The validator table confirms each analysis step passed its checks. Green means the step is sound.
Evidence Classification — 7 Dimensions
Dimension Classification Multiplier Contribution
Telemetry SINGLE 0.50 6.00
Dependency Mapping INFERRED 0.25 2.50
Procurement Contracts MISSING 0.00 0.00
Architecture Diagrams MISSING 0.00 0.00
Security Compliance INFERRED 0.25 1.25
Business Criticality SINGLE 0.50 2.00
Dr Requirements MISSING 0.00 0.00
Applied Cap: SRC-CAP-001
Validation Reports — Per Agent
Agent Status Provenance Counts Defaults Used Rule Failures
telemetry PASS
SRC:2CUST:10COMP:3DEF:1
 None ✓ None
dependency PASS
SRC:6CUST:5COMP:3DEF:2
 None ✓ None
procurement PASS
None ✓ None
provisioning PASS_WITH_DEFAULTS
SRC:7CUST:1COMP:9DEF:1
• {'default_id': 'DEF-WL-001', 'field': 'workload_type', 'value': 'app_server', 'r
• {'default_id': 'DEF-MIG-001', 'field': 'migration_factor', 'value': 0.2, 'reason
 ✓ None
synthesizer PASS
None ✓ None
confidence_advisor PASS
COMP:15
 None ✓ None
This section contains the full technical reasoning behind the recommendation: the rules that fired, the decision path, the confidence math, and the open intelligence gaps. It is intended for analysts and engineers. Executive readers can rely on the Overview tab.
Rules Applied
These are the specific governance rules that produced this recommendation.
▸ STRUCTURAL PRE-CHECK
This app was classified RETAIN before any 6R rule was evaluated, because it has hardware integration, latency sensitivity, or a life-safety classification that physically prevents cloud migration.
▸ SRC-CAP-001 (binding)
Confidence is capped at 40 because the app handles PHI and the vendor's Business Associate Agreement is not confirmed. PHI cannot move to a cloud vendor without a confirmed BAA.
▸ SRC-CAP-002
Confidence is capped at 55 because more than half of the seven evidence dimensions are inferred or missing — the conclusion is data-thin.
▸ SRC-CAP-003
Confidence is capped at 65 because none of the seven evidence dimensions reached CONFIRMED quality — there is no high-quality anchor for the analysis.
▸ SRC-CAP-004
Confidence is capped at 65 because telemetry is a single point-in-time snapshot with no ongoing monitoring — right-sizing needs usage patterns over time.
▸ RULE T2 — Right-Sizing Hard Cap
Right-sizing is limited to a moderate 25% reduction because telemetry is a single snapshot. More aggressive right-sizing needs 30+ days of monitoring data.
Decision Path
The step-by-step trace of how the recommendation was reached, in the system's own terms.
PRE-CHECK: hardware_dependency=UNKNOWN (not true) — pre-check does not fire. RULE 1: saas_available=false, saas_confidence_inferred=10 (<50) — does not fire. RULE 2: no decommission annotation, no named successor — does not fire. RULE 3: vendor COTS (Biotronik), not internally developed — does not fire. RULE 4: hardware_dependency=UNKNOWN, cannot confirm no_hardware_dependency=true — does not fire. RULE 5: life_safety_classification=true but hardware_dependency=UNKNOWN and latency_sensitive=UNKNOWN — partial match only. RESOLUTION: No rule fires cleanly. life_safety_classification=true + no confirmed migration path + vendor COTS with unconfirmed cloud compatibility → strategic_recommendation=RETAIN. CONFIDENCE: base=50, contributions=11.75, pre_cap=61.75. SRC-CAP-001 fires (phi_handled=true, baa_confirmed=UNKNOWN) → ceiling=40. SRC-CAP-002 fires (5/7 dimensions INFERRED/MISSING) → ceiling=55. SRC-CAP-003 fires (zero CONFIRMED) → ceiling=65. SRC-CAP-004 fires (snapshot_count=1, apm=false) → ceiling=65. Lowest cap: SRC-CAP-001 → final_confidence=40. GATE: strategic_recommendation=RETAIN → recommendation_status=STRUCTURAL, confidence_gate_applied=false. ROI VARIANCE: base=5 + telemetry=0(SINGLE) + procurement=15(MISSING) + arch=15(MISSING) + dep=10(INFERRED) + dr=10(MISSING) = 55% (below 60% cap).
Cap Evaluation (all fired caps): SRC-CAP-001: PHI handled + BAA not confirmed → ceiling 40 (ceiling=40) ← BINDING SRC-CAP-002: ≥4 of 7 dimensions INFERRED/MISSING → ceiling 55 (ceiling=55) SRC-CAP-003: Zero CONFIRMED dimensions → ceiling 65 (ceiling=65) SRC-CAP-004: Single snapshot, no APM → ceiling 65 (ceiling=65)
Confidence Scoring Breakdown
How the confidence percentage was calculated, dimension by dimension.
Base score: 50 Evidence contributions: Telemetry SINGLE +6.00 Dependency Mapping INFERRED +2.50 Procurement Contracts MISSING +0.00 Architecture Diagrams MISSING +0.00 Security Compliance INFERRED +1.25 Business Criticality SINGLE +2.00 Dr Requirements MISSING +0.00 Pre-cap score: 61.75 Binding cap applied: SRC-CAP-001 Final confidence: 40%
Intelligence Gaps
Specific things we don't yet know. Each gap is a reason the confidence score is held back.
hardware_dependency — UNKNOWN
Cannot confirm or deny structural RETAIN blocker. Biotronik PaceArt may require dedicated hardware interfaces (device programmers, serial/USB cardiac device readers). If hardware_dependency=true AND latency_sensitive=true, pre-check fires and RETAIN becomes unconditional. Requires Biotronik vendor confirmation and on-site infrastructure review.
BAA status unconfirmed — HIPAA compliance gap
baa_confirmed=UNKNOWN for a PHI-bearing life-safety system. SRC-CAP-001 caps confidence at 40. Any cloud migration path requires BAA execution before data transfer. Immediate legal/compliance review required regardless of cloud disposition.
EOL OS and database stack — active security risk
All 5 VMs run Windows Server 2012 R2 (EOL Oct 2023). Production DB runs SQL Server 2008 (ESU expired July 2022). Test DB runs SQL Server 2012 (EOL July 2022). PHI-bearing life-safety system running unpatched infrastructure is a HIPAA security rule violation risk. Immediate remediation required: Azure Arc ESU (on-prem) or OS/DB upgrade.
Biotronik cloud deployment compatibility — unconfirmed
No evidence that Biotronik supports PaceArt Optima in cloud-hosted environments. Vendor COTS medical device software frequently has OS/hardware certification constraints. Cloud migration cannot proceed without Biotronik written confirmation of supported deployment configurations.
latency_sensitive — UNKNOWN
Cardiac device interrogation workflows may have real-time data transfer requirements between device programmers and the PaceArt server. If latency_sensitive=true and hardware_dependency=true, pre-check fires unconditionally. Requires clinical workflow and network latency assessment.
DR requirements — MISSING
No DR documentation for a life-safety system. RTO/RPO unknown. Dual-datacenter deployment (MUSC-ART + MUSC-USB) suggests some HA awareness but no formal DR posture confirmed. Critical for HIPAA contingency planning requirements.
HL7/interface inventory — incomplete
epic_integration=false per catalog but PaceArt Optima supports HL7 interfaces. Downstream HL7 consumers (scheduling, billing, other EMR systems) not inventoried. Migration planning requires full interface map to avoid clinical workflow disruption.
Procurement contract and licensing terms — MISSING
contract_intelligence=INSUFFICIENT_EVIDENCE. Biotronik licensing terms for virtualized or cloud-hosted deployment unknown. Contract review required before any infrastructure change.
Technical Detail
A fuller technical explanation of the recommendation and its constraints.
PaceArt (Biotronik) is a vendor COTS cardiac device management system with life_safety_classification=true, phi_handled=true, and baa_confirmed=UNKNOWN. The pre-check structural blocker did not fire because hardware_dependency=UNKNOWN (not confirmed true), but no migration rule could fire: Rule 1 fails (saas_available=false, saas_confidence_inferred=10), Rule 2 fails (no decommission annotation or named successor), Rule 3 fails (vendor COTS, not internally developed), and Rule 4 fails (hardware_dependency=UNKNOWN prevents confirming no_hardware_dependency=true). RETAIN is the correct strategic disposition — not confidence-gate-forced, but structurally warranted given unresolved hardware dependency, life-safety classification, and absence of any confirmed migration path. SRC-CAP-001 drives confidence to 40 (phi_handled=true, baa_confirmed=UNKNOWN), compounded by SRC-CAP-002/003/004 from missing procurement, architecture, and DR evidence.
Secondary Path
The fallback option if the primary recommendation cannot proceed.
Secondary Path — REHOST (conditional — requires hardware_dependency=false confirmed by Biotronik)
REHOST becomes viable only if Biotronik confirms cloud-hosted deployment support, hardware_dependency is confirmed false, and a BAA is executed with the target cloud provider. Azure is the preferred secondary path given Azure Arc ESU availability for on-premises EOL OS remediation as an interim measure.
Right-Sizing Notes
Why the cloud sizing estimate is set where it is, based on the telemetry available.
Fleet avg memory active pct = 17.4%, which would ordinarily qualify for SRC-RS-003 (OVERSIZED, threshold <25%). However, snapshot_count=1 AND apm_telemetry_available=false trigger Rule T2 hard cap. Maximum allowed tier is SRC-RS-002 (MODERATE, 0.25 reduction); SRC-RS-003 (OVERSIZED) and SRC-RS-004 (ZOMBIE) are forbidden. Single-snapshot data cannot distinguish a genuinely idle VM from one sampled between bursts. CPU utilization data is entirely absent (cpu_avg_pct null on all 5 of 5 VMs). Rule T1 does not fire: application_type is Client/Server, not Hardware-Integrated. All 5 VMs run Windows Server 2012 R2 (EOL October 2023), constituting active HIPAA Security Rule risk given phi_handling=true.
This tab shows the raw output of each analysis agent — the specialized components that examine telemetry, dependencies, contracts, and cost. It is the most detailed and most technical view. Most readers will find the Overview and Backlog tabs sufficient.
Telemetry Agent
Utilization Profilemoderate
Snapshot Count1
APM AvailableFalse
Hard Rule TriggeredTrue
Right-Sizing Opp.True
Agent Confidence47%
Telemetry — VM Counts
Powered Off VMs0
Flagged VMs0
Mem Active % (avg)17.4%
Data WindowNone
Inventory Source
Dependency Agent
Epic IntegrationFalse
PHI HandledTrue
Hardware DependencyUNKNOWN
Latency SensitiveUNKNOWN
Blast Radiuscritical
Migration RiskHIGH
Conf (Inferred)72%
Conf (Confirmed)41%
Verificationsingle_source
Cloud Connectivity
ComplianceHIPAA / HITRUST
Upstream (4)
SQL Server 2008 (optimadbprod): confirmed: rvtools.vm_notes — 'Windows Server 2012 R2 w/SQL 2008 Larry Norman: PaceArt'
SQL Server 2012 (optimadbtst): confirmed: rvtools.vm_notes — 'Windows Server 2012 R2 w/SQL 2012 Larry Norman: PaceArt'
Windows Server 2012 R2 (all 5 VMs): confirmed: rvtools.vm_notes — all five VMs note 'Windows Server 2012 R2'
Biotronik PaceArt Optima application stack: confirmed: catalog.vendor = 'Biotronik', catalog.application_name = 'PaceArt', rvtools.vm_notes corroborate across all…
Downstream (2)
Cardiac Device Follow-Up Clinical Workflows: inferred: application domain — PaceArt is used by cardiac device clinic staff for pacemaker/ICD follow-up scheduling,…
Potential Epic integration (unconfirmed): inferred: industry pattern — PaceArt Optima supports HL7 interfaces to EMR systems; catalog.epic_integration=false but…
Procurement Agent
SaaS Available—
SaaS Vendor
SaaS Product
Repurchase ViableFalse
BAA ConfirmedUNKNOWN
SaaS Confidence
Inferred Score10%
Confirmed Score0%
Verificationinferred_only
AWS Equiv.None
Provisioning Agent (Python engine — not LLM)
Current VMs5
Recommended VMs4
Right-Sizing25% reduction
Current TCO/yr$32,500
Projected Cloud TCO$20,800
Migration One-Time$4,160
Dual-Running Cost$17,767
N/A — RETAIN
Year-1 Net
N/A — RETAIN
Steady-State
N/A — RETAIN
3-Year
Migration cost figures are not computed for structurally retained applications.
AWS TargetAmazon EC2
Azure TargetAzure Virtual Machines
Cost Modeldefault_multipliers
DEF: {'default_id': 'DEF-WL-001', 'field': 'workload_type', 'value': 'app_server', 'reason': 'workload_type indeterminable'}
DEF: {'default_id': 'DEF-MIG-001', 'field': 'migration_factor', 'value': 0.2, 'reason': 'migration_complexity not provided'}
DEF: {'default_id': 'DEF-COST-001', 'field': 'cost_model_basis', 'value': 'default_multipliers', 'reason': 'no customer cloud quote provided; using SRC-TCO-* and SRC-MULT-* registry defaults'}
GE Centricity / CCW
Vendor: GE Healthcare · 12 VMs · Criticality: High · HIPAA / HITRUST — phi_handled=true; BAA
REPURCHASE
40%
Confidence
32%
Data Quality
ROI: Year-1 Net: -$19,760 · Steady-State: $31,200/yr · 3-Year: $42,640
Overview
Backlog
Provenance
Technical Details
Agents
Recommendation
REPURCHASE ⏳ PENDING CONDITIONS
SRC-CAP-001 Binding cap · Ceiling: 40% · Threshold: 60%
Secondary: REHOST
Confidence
40%
Viable, but specific gaps must close first.
ROI Summary (Python-computed, not LLM)
-$19,760
Year-1 Net
$31,200
Steady-State /yr
$42,640
3-Year Total
ROI variance: ±60%
⚙ Financials computed by provisioning_engine_v3 (Python) — not LLM-generated
In plain terms: migrating costs money up front — Year 1 is a net outflow of $19,760. From Year 2 onward the app saves $31,200 every year. The 3-Year total nets these against each other.
Finding
GE Centricity / CCW has a confirmed vendor SaaS successor — GE Centricity Cloud and the GE Edison Digital Health Platform — making it a strong REPURCHASE candidate that would eliminate 12 on-premises Windows Server VMs, including one running end-of-life Windows Server 2008 R2. However, this workload handles protected health information and the Business Associate Agreement status with GE Healthcare is currently unconfirmed, which is a compliance prerequisite before any migration can proceed. Resolving the BAA, validating SaaS contract terms, and mapping downstream clinical consumers are the three conditions required to unlock this recommendation.
In plain terms: this app should move to the vendor's cloud product, but a few specific items must be confirmed first. See Conditions to Unlock below.
Conditions to Unlock (2 items)
This recommendation becomes executable once these conditions are confirmed:
☐ Verify the Business Associate Agreement with GE Healthcare before any migration can proceed
Owner: Compliance Officer / Legal Counsel · Timeline: 21 days · Unlocks: compliance_close, cap_resolution, migration_commit
☐ Get formal pricing and contract terms from GE Healthcare for the cloud version of Centricity
Owner: Procurement / Vendor Management · Timeline: 21 days · Unlocks: cost_clarity, migration_commit
See Backlog tab for full action backlog and P2/P3 items.
This is the list of work needed to move this recommendation forward. The 2 P1 items are the priority — they unlock the recommendation. P2 and P3 items improve confidence and reduce risk but are not blockers. 'Story points' measure effort, not time or cost — see the Methodology tab for how they work.
40%
Current
60%
Threshold
55%
Ceiling
29
Story Points
Why This Score
Final confidence is hard-capped at 40 by SRC-CAP-001 because phi_handled=true and baa_confirmed=UNKNOWN — a PHI-bearing workload with no confirmed Business Associate Agreement cannot exceed this ceiling regardless of other evidence quality. The pre-cap score of 66.0 reflects meaningful catalog and RVTools signal, but three additional caps (SRC-CAP-002 at 55, SRC-CAP-003 at 65, SRC-CAP-004 at 65) form a sequential constraint chain that will govern confidence growth after SRC-CAP-001 is resolved.
Proceeding with REPURCHASE planning without a confirmed BAA exposes the organization to HIPAA enforcement risk and contractual liability, while the concurrent on-premises upgrade project (PRJTASK0107612) and an active EOL OS on CCEIVDATACONV create compounding execution risk if migration is initiated before these blockers are resolved.
Action Backlog — 6 Items · 29 Story Points · Ceiling: 55%
APP-004-001 Confirm BAA Status with GE Healthcare for SaaS Platform
P1 BAA/Legal
Retrieve or execute a Business Associate Agreement (BAA) with GE Healthcare covering the GE Centricity Cloud / GE Edison Digital Health Platform SaaS offering. The current catalog field baa_valid=null means baa_confirmed=UNKNOWN under v3.1 tri-state rules, which triggers SRC-CAP-001 and hard-caps confidence at 40. This is the single binding constraint preventing the REPURCHASE recommendation from
+15% confidence -0% ROI variance 👤 📋 · 8 pts
APP-004-002 Obtain GE Healthcare SaaS Contract Terms, Pricing, and Migration Proposal
P1 Procurement
Engage GE Healthcare procurement to obtain: (1) a customer-specific RFP response or migration proposal for GE Centricity Cloud / GE Edison Digital Health Platform, (2) ACV and contract renewal date for the existing on-premises Centricity license, and (3) total cost of ownership delta between on-premises and SaaS. Current state: contract_acv=null, contract_renewal_date=null, saas_confidence_confirm
+0% confidence -0% ROI variance 👤 📋 · 8 pts
APP-004-003 Map Downstream Clinical Consumers of GE Centricity / CCW
P2 Dependency Mapping
Conduct a formal interface inventory to enumerate all downstream systems consuming GE Centricity / CCW result feeds, imaging data, and HL7/DICOM outputs. Current state: downstream_dependencies are inferred from product class only — no catalog or RVTools artifact names specific EHR result feeds, ordering systems, or departmental viewers. Blast radius is held at 'medium' pending owner confirmation o
+0% confidence -0% ROI variance 👤 📋 · 5 pts
APP-004-004 Deploy APM Telemetry and Collect 30-Day Utilization Baseline
P2 Telemetry
Deploy application performance monitoring (APM) tooling across the 11 powered-on GE Centricity / CCW VMs and collect a minimum 30-day utilization baseline covering CPU, memory, disk I/O, and network throughput. Current state: apm_telemetry_available=false, snapshot_count=1, fleet_avg_cpu_pct=null for all 12 VMs. SRC-RS-002 hard rule caps right-sizing at MODERATE (25% reduction) due to single-snaps
+0% confidence -0% ROI variance 👤 📋 · 3 pts
APP-004-005 Remediate End-of-Life Windows Server 2008 R2 on CCEIVDATACONV
P2 Security/Compliance
VM CCEIVDATACONV is running Windows Server 2008 R2, which reached end of extended support on January 14, 2020. This VM is powered ON in production (MUSC-USB datacenter) and is annotated as a 'GE Centricity Migration Server'. Operating an unsupported OS in a PHI-bearing environment constitutes an active HIPAA Security Rule violation under 45 CFR §164.312 (Technical Safeguards) — compliance flag HIP
+0% confidence -0% ROI variance 👤 📋 · 3 pts
APP-004-006 Document DR Requirements and Validate RTO/RPO Against SaaS Platform Capabilities
P3 Architecture
No DR documentation is cited across any agent output for this High-criticality PHI workload. Validated RTO/RPO targets are required to: (1) assess whether GE Centricity Cloud SaaS DR capabilities meet current on-premises posture, (2) determine whether the multi-datacenter topology (MUSC-USB primary, MUSC-WAN secondary) implies an active-passive DR configuration that must be replicated in the SaaS
+0% confidence -0% ROI variance 👤 📋 · 2 pts
This tab shows the evidence behind the recommendation. The seven dimensions are scored by quality — CONFIRMED evidence counts fully, INFERRED evidence counts less. The validator table confirms each analysis step passed its checks. Green means the step is sound.
Evidence Classification — 7 Dimensions
Dimension Classification Multiplier Contribution
Telemetry SINGLE 0.50 6.00
Dependency Mapping SINGLE 0.50 5.00
Procurement Contracts MISSING 0.00 0.00
Architecture Diagrams INFERRED 0.25 1.75
Security Compliance INFERRED 0.25 1.25
Business Criticality SINGLE 0.50 2.00
Dr Requirements MISSING 0.00 0.00
Applied Cap: SRC-CAP-001
Validation Reports — Per Agent
Agent Status Provenance Counts Defaults Used Rule Failures
telemetry PASS
SRC:2CUST:8COMP:4DEF:1
 None ✓ None
dependency PASS
CUST:2COMP:8DEF:3
 None ✓ None
procurement PASS
None ✓ None
provisioning PASS_WITH_DEFAULTS
SRC:7CUST:1COMP:9DEF:1
• {'default_id': 'DEF-WL-001', 'field': 'workload_type', 'value': 'app_server', 'r
• {'default_id': 'DEF-MIG-001', 'field': 'migration_factor', 'value': 0.2, 'reason
 ✓ None
synthesizer PASS
None ✓ None
confidence_advisor PASS
COMP:15
 None ✓ None
This section contains the full technical reasoning behind the recommendation: the rules that fired, the decision path, the confidence math, and the open intelligence gaps. It is intended for analysts and engineers. Executive readers can rely on the Overview tab.
Rules Applied
These are the specific governance rules that produced this recommendation.
▸ RULE 1 — REPURCHASE
A vendor SaaS replacement is available and credible, and the app has no hardware constraint — so the recommendation is to replace it with the vendor's cloud service.
▸ SRC-CAP-001 (binding)
Confidence is capped at 40 because the app handles PHI and the vendor's Business Associate Agreement is not confirmed. PHI cannot move to a cloud vendor without a confirmed BAA.
▸ SRC-CAP-002
Confidence is capped at 55 because more than half of the seven evidence dimensions are inferred or missing — the conclusion is data-thin.
▸ SRC-CAP-003
Confidence is capped at 65 because none of the seven evidence dimensions reached CONFIRMED quality — there is no high-quality anchor for the analysis.
▸ SRC-CAP-004
Confidence is capped at 65 because telemetry is a single point-in-time snapshot with no ongoing monitoring — right-sizing needs usage patterns over time.
▸ RULE T2 — Right-Sizing Hard Cap
Right-sizing is limited to a moderate 25% reduction because telemetry is a single snapshot. More aggressive right-sizing needs 30+ days of monitoring data.
Decision Path
The step-by-step trace of how the recommendation was reached, in the system's own terms.
PRE-CHECK: hardware_dependency=false AND life_safety_classification=false → pre-check does not fire. RULE 1 (REPURCHASE): saas_alternative_available=true (catalog.notes confirmed) AND saas_confidence_inferred=70 >= 50 AND hardware_dependency=false → RULE 1 FIRES → strategic_recommendation=REPURCHASE. GATE EVALUATION: final_confidence=40 < 60 AND strategic_recommendation=REPURCHASE (not RETAIN) → recommendation_status=PENDING_CONDITIONS, displayed_primary=REPURCHASE (v4.1: gate does not force RETAIN), confidence_gate_applied=true. CAP TRACE: pre-cap=66.0; SRC-CAP-001 fires (phi_handled=true, baa_confirmed=UNKNOWN) → cap 40; SRC-CAP-002 fires (4/7 dimensions INFERRED/MISSING) → cap 55; SRC-CAP-003 fires (zero CONFIRMED dimensions) → cap 65; SRC-CAP-004 fires (snapshot_count=1, apm_telemetry_available=false) → cap 65; lowest cap = SRC-CAP-001 at 40 → final_confidence=40.
Cap Evaluation (all fired caps): SRC-CAP-001: PHI handled + BAA not confirmed → ceiling 40 (ceiling=40) ← BINDING SRC-CAP-002: ≥4 of 7 dimensions INFERRED/MISSING → ceiling 55 (ceiling=55) SRC-CAP-003: Zero CONFIRMED dimensions → ceiling 65 (ceiling=65) SRC-CAP-004: Single snapshot, no APM → ceiling 65 (ceiling=65)
Confidence Scoring Breakdown
How the confidence percentage was calculated, dimension by dimension.
Base score: 50 Evidence contributions: Telemetry SINGLE +6.00 Dependency Mapping SINGLE +5.00 Procurement Contracts MISSING +0.00 Architecture Diagrams INFERRED +1.75 Security Compliance INFERRED +1.25 Business Criticality SINGLE +2.00 Dr Requirements MISSING +0.00 Pre-cap score: 66.0 Binding cap applied: SRC-CAP-001 Final confidence: 40%
Intelligence Gaps
Specific things we don't yet know. Each gap is a reason the confidence score is held back.
BAA Status with GE Healthcare SaaS Platform Unconfirmed
phi_handled=true workload cannot migrate to GE Centricity Cloud without a validated Business Associate Agreement. This is the binding compliance prerequisite and the primary driver of SRC-CAP-001 (confidence ceiling 40). Resolution unlocks the REPURCHASE path to EXECUTABLE status.
SaaS Contract Terms and Pricing Not Available
contract_intelligence=INSUFFICIENT_EVIDENCE and saas_confidence_confirmed=0 mean ROI projection carries 60% variance. Procurement engagement with GE Healthcare required to validate SaaS licensing model, migration support terms, and total cost of ownership delta.
Downstream Clinical Consumer Mapping Incomplete
Downstream dependencies (EHR result feeds, ordering systems, departmental viewers) are inferred from product class, not confirmed from catalog or RVTools artifacts. Blast radius estimate is 'medium' but unvalidated. Interface inventory required before cutover planning.
APM Telemetry Absent — Right-Sizing Confidence Limited
apm_telemetry_available=false with snapshot_count=1 limits right-sizing confidence to SRC-RS-002 (MODERATE, 25% reduction). Application-layer performance data needed to validate the 9-VM post-migration fleet size and confirm no workload consolidation risk.
End-of-Life OS on CCEIVDATACONV (Windows Server 2008 R2)
One VM is running Windows Server 2008 R2, which reached end of extended support in January 2020. This represents an active security and compliance risk under HIPAA/HITRUST. Immediate remediation or accelerated migration timeline warranted regardless of SaaS transition status.
DR and RTO/RPO Requirements Not Documented
No DR documentation cited across any agent output. High-criticality PHI workload requires validated RTO/RPO targets to assess SaaS platform DR capabilities against current on-premises posture.
Architecture Diagrams Absent
Application architecture inferred from VM naming patterns and product class only. No formal architecture diagrams available to validate integration topology, data flows, or migration sequencing requirements.
Technical Detail
A fuller technical explanation of the recommendation and its constraints.
Procurement agent confirmed saas_available=true with saas_confidence_inferred=70, referencing catalog.notes citing GE Centricity Cloud / GE Edison Digital Health Platform as the vendor SaaS successor — satisfying all three conditions for Rule 1 (REPURCHASE): SaaS available, inferred confidence ≥ 50, and no hardware dependency. Dependency agent confirmed hardware_dependency=false and life_safety_classification=false, clearing the structural pre-check. The binding constraint is SRC-CAP-001: phi_handled=true with baa_confirmed=UNKNOWN drives the confidence ceiling to 40, placing this recommendation in PENDING_CONDITIONS status. The secondary recommendation is REHOST as a lower-risk interim path if SaaS transition timelines extend, given the VM-based architecture with no hardware constraints.
Secondary Path
The fallback option if the primary recommendation cannot proceed.
Secondary Path — REHOST
REHOST is viable as an interim disposition: 12 Windows Server VMs with no hardware dependency and no latency constraint are straightforward IaaS candidates. This path preserves optionality while SaaS contract and BAA negotiations proceed, and eliminates the end-of-life Windows Server 2008 R2 exposure on CCEIVDATACONV.
Right-Sizing Notes
Why the cloud sizing estimate is set where it is, based on the telemetry available.
Single point-in-time RVTools snapshot with no APM telemetry. Hard rule T2 (SRC-RS-002) caps maximum tier at MODERATE (0.25 reduction); OVERSIZED and ZOMBIE tiers are forbidden regardless of observed utilization. Fleet avg mem active pct is 25.2% across 11 powered-on VMs, which under Rule T3 would independently yield MODERATE (>=25% threshold). CPU utilization data is absent for all 12 VMs; fleet_avg_cpu_pct cannot be computed. One VM (CCEIVDATACONV) runs Windows Server 2008 R2 (EOL January 2020), constituting an active HIPAA Security Rule risk given phi_handling=true. One VM (KMC-CPN-LS) is powered off. Application type is Client/Server with no life_safety_classification present; Rule T1 override does not apply.
This tab shows the raw output of each analysis agent — the specialized components that examine telemetry, dependencies, contracts, and cost. It is the most detailed and most technical view. Most readers will find the Overview and Backlog tabs sufficient.
Telemetry Agent
Utilization Profilemoderate
Snapshot Count1
APM AvailableFalse
Hard Rule TriggeredTrue
Right-Sizing Opp.True
Agent Confidence48%
Telemetry — VM Counts
Powered Off VMs1
Flagged VMs0
Mem Active % (avg)25.2%
Data Window1 days
Inventory Source
Dependency Agent
Epic IntegrationFalse
PHI HandledTrue
Hardware DependencyFalse
Latency SensitiveFalse
Blast Radiusmedium
Migration RiskMEDIUM
Conf (Inferred)72%
Conf (Confirmed)55%
Verificationsingle_source
Cloud Connectivity
ComplianceHIPAA / HITRUST
Upstream (3)
GE Centricity Migration / Data Conversion Service: confirmed: rvtools.vm_notes — CCEIVDATACONV note: 'GE Centricity Migration Server'
GE CCW Integration Layer (Production): confirmed: rvtools.vm_notes — geccwintprod / geccwintcolprd notes reference CCW integration VMs with MAC address…
GE CCW Database (Production): confirmed: rvtools.vm_notes — geccwdbprod note: 'GE/CCW Upgrade > Build'; geccwnewproddb note: same project task
Downstream (2)
Clinical Consumers of CCW Results / Imaging Data (unspecified): inferred: GE Centricity / CCW is a results and imaging management platform; downstream clinical consumers (EHR result…
GE Edison Digital Health Platform / GE Centricity Cloud (SaaS successor): confirmed: catalog.notes — 'SaaS alternative: GE Centricity Cloud / GE Edison Digital Health Platform.'
Procurement Agent
SaaS Available—
SaaS VendorGE Healthcare
SaaS ProductGE Centricity Cloud / GE Edison Digital Health Platform
Repurchase ViableINSUFFICIENT_EVIDENCE
BAA ConfirmedUNKNOWN
SaaS Confidence
Inferred Score70%
Confirmed Score0%
Verificationsingle_source
AWS Equiv.None
Provisioning Agent (Python engine — not LLM)
Current VMs12
Recommended VMs9
Right-Sizing25% reduction
Current TCO/yr$78,000
Projected Cloud TCO$46,800
Migration One-Time$9,360
Dual-Running Cost$41,600
-$19,760
Year-1 Net
$31,200
Steady-State
$42,640
3-Year
AWS TargetAmazon EC2
Azure TargetAzure Virtual Machines
Cost Modeldefault_multipliers
DEF: {'default_id': 'DEF-WL-001', 'field': 'workload_type', 'value': 'app_server', 'reason': 'workload_type indeterminable'}
DEF: {'default_id': 'DEF-MIG-001', 'field': 'migration_factor', 'value': 0.2, 'reason': 'migration_complexity not provided'}
DEF: {'default_id': 'DEF-COST-001', 'field': 'cost_model_basis', 'value': 'default_multipliers', 'reason': 'no customer cloud quote provided; using SRC-TCO-* and SRC-MULT-* registry defaults'}
Mosaiq
Vendor: Elekta · 3 VMs · Criticality: High · HIPAA / HITRUST; phi_handled=true; life_
REPURCHASE
40%
Confidence
32%
Data Quality
ROI: Year-1 Net: -$2,947 · Steady-State: $9,100/yr · 3-Year: $15,253
Overview
Backlog
Provenance
Technical Details
Agents
Recommendation
REPURCHASE ⏳ PENDING CONDITIONS
SRC-CAP-001 Binding cap · Ceiling: 40% · Threshold: 60%
Secondary: RETAIN
Confidence
40%
Viable, but specific gaps must close first.
ROI Summary (Python-computed, not LLM)
-$2,947
Year-1 Net
$9,100
Steady-State /yr
$15,253
3-Year Total
ROI variance: ±45%
⚙ Financials computed by provisioning_engine_v3 (Python) — not LLM-generated
In plain terms: migrating costs money up front — Year 1 is a net outflow of $2,947. From Year 2 onward the app saves $9,100 every year. The 3-Year total nets these against each other.
Finding
Mosaiq 2.6 is a high-criticality radiation oncology information system handling protected health information across three on-premises VMs. A vendor-native SaaS path exists via Elekta MOSAIQ Online, making REPURCHASE the strategically preferred disposition — however, three critical conditions must be resolved before this recommendation can be executed: the BAA with Elekta must be confirmed, the hardware integration between the ESI server and the linear accelerator must be formally documented, and latency requirements for treatment delivery workflows must be assessed. Until these conditions are met, the recommendation is held at PENDING_CONDITIONS.
In plain terms: this app should move to the vendor's cloud product, but a few specific items must be confirmed first. See Conditions to Unlock below.
Conditions to Unlock (3 items)
This recommendation becomes executable once these conditions are confirmed:
☐ Obtain Signed Business Associate Agreement with Elekta for Cloud Product
Owner: Compliance Officer / Legal Counsel · Timeline: 10 days · Unlocks: cap_resolution, compliance_close, migration_commit
☐ Confirm Whether Mosaiq Is Physically Wired to the Linear Accelerator
Owner: RAD/ONC Clinical Engineering Lead (Jim Early / Nayquan Scott) · Timeline: 21 days · Unlocks: migration_commit, operational_safety
☐ Determine Whether Real-Time Speed Is Required for Radiation Treatment Delivery
Owner: RAD/ONC Physics / Clinical Engineering (Jim Early / Nayquan Scott) · Timeline: 10 days · Unlocks: migration_commit, operational_safety
See Backlog tab for full action backlog and P2/P3 items.
This is the list of work needed to move this recommendation forward. The 3 P1 items are the priority — they unlock the recommendation. P2 and P3 items improve confidence and reduce risk but are not blockers. 'Story points' measure effort, not time or cost — see the Methodology tab for how they work.
40%
Current
60%
Threshold
55%
Ceiling
15
Story Points
Why This Score
Confidence is capped at 40 by SRC-CAP-001 (phi_handled=true, baa_confirmed=UNKNOWN) — the binding constraint that prevents any confidence climb regardless of other evidence quality. The pre-cap score of 64.25 demonstrates that the underlying application intelligence is substantive, but the BAA gap alone is sufficient to hold the recommendation in PENDING_CONDITIONS status until legal confirmation is obtained.
Executing a REPURCHASE migration of a life-safety radiation oncology system to cloud without a confirmed BAA, unresolved Linac hardware dependency, and zero confirmed telemetry dimensions exposes the organization to simultaneous HIPAA legal liability, potential clinical safety risk, and a migration that may need to be reversed if hardware_dependency is confirmed true.
Action Backlog — 6 Items · 15 Story Points · Ceiling: 55%
APP-005-001 Confirm BAA with Elekta for MOSAIQ Online SaaS — Resolve SRC-CAP-001
P1 BAA/Legal
Engage Elekta's legal and compliance team to obtain a signed Business Associate Agreement (BAA) covering MOSAIQ Online (SaaS) for your organization as a covered entity. The current catalog field baa_valid=null means BAA status is UNKNOWN, which triggers SRC-CAP-001 and caps confidence at 40 — the single highest-impact action in this backlog. A signed BAA is also a HIPAA legal prerequisite before a
+15% confidence -0% ROI variance 👤 📋 · 2 pts
APP-005-002 Verify Linac Hardware Dependency via Elekta ESI Integration Documentation
P1 Dependency Mapping
The msqesi VM is identified as 'external system interface server RAD/ONC' — the standard Elekta integration point for linear accelerator (Linac) hardware. hardware_dependency is currently UNKNOWN because catalog.application_type='Client/Server' conflicts with the ESI VM's implied physical coupling role. This is a binary decision gate: if hardware_dependency is confirmed true alongside life_safety_
+0% confidence -0% ROI variance 👤 📋 · 3 pts
APP-005-003 Assess Latency Requirements for Mosaiq ESI Treatment Delivery Interface
P1 Architecture
latency_sensitive=UNKNOWN for Mosaiq's ESI treatment delivery interface. Radiation therapy delivery sequencing may impose real-time or near-real-time latency requirements on the Linac control pathway. If latency_sensitive is confirmed true alongside hardware_dependency=true, the structural pre-check fires and the recommendation must be revised to RETAIN. Even if hardware_dependency=false, latency
+0% confidence -0% ROI variance 👤 📋 · 2 pts
APP-005-004 Engage Elekta for MOSAIQ Online Pricing, SLA, and Migration Scope
P2 Procurement
saas_confidence_confirmed=0 and contract_intelligence=INSUFFICIENT_EVIDENCE. The current ROI projection of $9,100 annual savings is based entirely on default VM multipliers and does not reflect actual MOSAIQ Online subscription costs, data migration fees, or clinical validation costs. repurchase_viable=INSUFFICIENT_EVIDENCE cannot be elevated to true without a customer-specific RFP response or exe
+0% confidence -0% ROI variance 👤 📋 · 3 pts
APP-005-005 Deploy 30-Day APM Telemetry Collection Across All Three Mosaiq VMs
P2 Telemetry
snapshot_count=1 and apm_telemetry_available=false, triggering SRC-CAP-004 (ceiling 65) and hard rule T2 which caps right-sizing at MODERATE (SRC-RS-002, 25% reduction). CPU utilization is null across all three VMs (msqapp, msqesi, msqsql). Fleet avg mem_active_pct=18% suggests potential OVERSIZED classification, but this cannot be confirmed from a single snapshot — a single point-in-time reading
+0% confidence -0% ROI variance 👤 📋 · 3 pts
APP-005-006 Document RTO/RPO Requirements and Validate Against MOSAIQ Online SLA
P3 Architecture
dr_requirements dimension is MISSING. As a life-safety classified radiation oncology system, Mosaiq requires formally documented RTO (Recovery Time Objective) and RPO (Recovery Point Objective) targets approved by clinical leadership. These must be validated against Elekta MOSAIQ Online SLA commitments before any SaaS migration can be approved. Engage clinical leadership, RAD/ONC department head,
+0% confidence -0% ROI variance 👤 📋 · 2 pts
This tab shows the evidence behind the recommendation. The seven dimensions are scored by quality — CONFIRMED evidence counts fully, INFERRED evidence counts less. The validator table confirms each analysis step passed its checks. Green means the step is sound.
Evidence Classification — 7 Dimensions
Dimension Classification Multiplier Contribution
Telemetry SINGLE 0.50 6.00
Dependency Mapping SINGLE 0.50 5.00
Procurement Contracts MISSING 0.00 0.00
Architecture Diagrams MISSING 0.00 0.00
Security Compliance INFERRED 0.25 1.25
Business Criticality SINGLE 0.50 2.00
Dr Requirements MISSING 0.00 0.00
Applied Cap: SRC-CAP-001
Validation Reports — Per Agent
Agent Status Provenance Counts Defaults Used Rule Failures
telemetry PASS
SRC:2CUST:9COMP:4DEF:2
 None ✓ None
dependency PASS
SRC:4CUST:6COMP:2DEF:1
 None ✓ None
procurement PASS
None ✓ None
provisioning PASS_WITH_DEFAULTS
SRC:7CUST:1COMP:9DEF:1
• {'default_id': 'DEF-WL-001', 'field': 'workload_type', 'value': 'app_server', 'r
• {'default_id': 'DEF-MIG-001', 'field': 'migration_factor', 'value': 0.2, 'reason
 ✓ None
synthesizer PASS
None ✓ None
confidence_advisor PASS
COMP:15
 None ✓ None
This section contains the full technical reasoning behind the recommendation: the rules that fired, the decision path, the confidence math, and the open intelligence gaps. It is intended for analysts and engineers. Executive readers can rely on the Overview tab.
Rules Applied
These are the specific governance rules that produced this recommendation.
▸ RULE 1 — REPURCHASE
A vendor SaaS replacement is available and credible — so the recommendation is to replace it with the vendor's cloud service.
▸ SRC-CAP-001 (binding)
Confidence is capped at 40 because the app handles PHI and the vendor's Business Associate Agreement is not confirmed. PHI cannot move to a cloud vendor without a confirmed BAA.
▸ SRC-CAP-002
Confidence is capped at 55 because more than half of the seven evidence dimensions are inferred or missing — the conclusion is data-thin.
▸ SRC-CAP-003
Confidence is capped at 65 because none of the seven evidence dimensions reached CONFIRMED quality — there is no high-quality anchor for the analysis.
▸ SRC-CAP-004
Confidence is capped at 65 because telemetry is a single point-in-time snapshot with no ongoing monitoring — right-sizing needs usage patterns over time.
▸ RULE T2 — Right-Sizing Hard Cap
Right-sizing is limited to a moderate 25% reduction because telemetry is a single snapshot. More aggressive right-sizing needs 30+ days of monitoring data.
Decision Path
The step-by-step trace of how the recommendation was reached, in the system's own terms.
PRE-CHECK: hardware_dependency=UNKNOWN (not true) → pre-check does not fire. RULE 1: saas_alternative_available=true (procurement.saas_available=true, SOURCED from catalog notes + procurement agent); saas_confidence_inferred=70 ≥ 50; hardware_dependency=UNKNOWN ≠ true → Rule 1 fires → strategic_recommendation=REPURCHASE. CONFIDENCE SCORING: base=50; telemetry=SINGLE(+6.0); dependency_mapping=SINGLE(+5.0); procurement_contracts=MISSING(+0); architecture_diagrams=MISSING(+0); security_compliance=INFERRED(+1.25); business_criticality=SINGLE(+2.0); dr_requirements=MISSING(+0); pre_cap_total=64.25. CAP EVALUATION: SRC-CAP-001 fires (phi_handled=true, baa_confirmed=UNKNOWN≠true) → cap=40; SRC-CAP-002 fires (4/7 dimensions MISSING/INFERRED = majority) → cap=55; SRC-CAP-003 fires (zero CONFIRMED dimensions) → cap=65; SRC-CAP-004 fires (snapshot_count=1, apm_telemetry_available=false) → cap=65. Lowest cap wins: SRC-CAP-001=40. final_confidence=40. GATE: final_confidence=40 < 60, strategic_recommendation=REPURCHASE (not RETAIN) → recommendation_status=PENDING_CONDITIONS; confidence_gate_applied=true; displayed_primary=REPURCHASE (v4.1: gate does not force RETAIN).
Cap Evaluation (all fired caps): SRC-CAP-001: PHI handled + BAA not confirmed → ceiling 40 (ceiling=40) ← BINDING SRC-CAP-002: ≥4 of 7 dimensions INFERRED/MISSING → ceiling 55 (ceiling=55) SRC-CAP-003: Zero CONFIRMED dimensions → ceiling 65 (ceiling=65) SRC-CAP-004: Single snapshot, no APM → ceiling 65 (ceiling=65)
Confidence Scoring Breakdown
How the confidence percentage was calculated, dimension by dimension.
Base score: 50 Evidence contributions: Telemetry SINGLE +6.00 Dependency Mapping SINGLE +5.00 Procurement Contracts MISSING +0.00 Architecture Diagrams MISSING +0.00 Security Compliance INFERRED +1.25 Business Criticality SINGLE +2.00 Dr Requirements MISSING +0.00 Pre-cap score: 64.25 Binding cap applied: SRC-CAP-001 Final confidence: 40%
Intelligence Gaps
Specific things we don't yet know. Each gap is a reason the confidence score is held back.
BAA Confirmation with Elekta MOSAIQ Online
phi_handled=true and baa_confirmed=UNKNOWN triggers SRC-CAP-001, capping confidence at 40 and blocking EXECUTABLE status. A signed BAA with Elekta for the SaaS product is a legal prerequisite under HIPAA before any PHI can be processed in the cloud environment. Resolution unlocks SRC-CAP-001 and raises confidence ceiling to 55+.
Hardware Dependency Verification — Linac ESI Integration
hardware_dependency=UNKNOWN. The msqesi VM is identified as 'external system interface server RAD/ONC' — the standard Elekta integration point for linear accelerator hardware. If hardware_dependency is confirmed true, the structural pre-check fires (hardware_dependency=true AND life_safety_classification=true) and the recommendation must be revised from REPURCHASE to RETAIN. Requires review of Elekta integration documentation and confirmation from Jim Early / Nayquan Scott (RAD/ONC POC).
Latency Sensitivity Assessment for Treatment Delivery Workflows
latency_sensitive=UNKNOWN. Radiation therapy treatment delivery may impose real-time or near-real-time latency requirements on the Mosaiq ESI interface. If latency_sensitive is confirmed true alongside hardware_dependency=true, the structural pre-check fires. Clinical workflow assessment required from RAD/ONC physics and IT teams.
SaaS Contract Intelligence — Elekta MOSAIQ Online
contract_intelligence=INSUFFICIENT_EVIDENCE; saas_confidence_confirmed=0. No pricing, SLA, migration timeline, or data migration scope has been obtained from Elekta. ROI projection of $9,100 annual savings is based on default VM multipliers and does not reflect actual SaaS subscription costs. Procurement engagement with Elekta required.
Architecture Documentation — Integration Topology
No architecture diagrams available. The three-VM topology (app server, ESI, SQL) and its integration with downstream RAD/ONC clinical workflows is inferred from VM notes only. Full integration map required to assess migration complexity and data migration scope for MOSAIQ Online.
DR Requirements for Life-Safety Clinical System
dr_requirements dimension is MISSING. As a life-safety classified system, Mosaiq requires documented RTO/RPO targets. These must be validated against Elekta MOSAIQ Online SLA commitments before any SaaS migration can be approved by clinical leadership.
Technical Detail
A fuller technical explanation of the recommendation and its constraints.
The dependency agent identified life_safety_classification=true and inferred a Linac hardware dependency via the ESI VM (msqesi: 'external system interface server RAD/ONC'), but hardware_dependency remains UNKNOWN — not confirmed true — so the structural pre-check does not fire. The procurement agent confirmed saas_available=true with Elekta MOSAIQ Online at saas_confidence_inferred=70, satisfying Rule 1's threshold of ≥50, and hardware_dependency=UNKNOWN does not equal true, so the Rule 1 hardware blocker does not apply. REPURCHASE is therefore the strategic recommendation. However, SRC-CAP-001 applies because phi_handled=true and baa_confirmed=UNKNOWN, capping confidence at 40 — below the 60-point execution threshold — placing the recommendation in PENDING_CONDITIONS status. The secondary recommendation is RETAIN, which becomes operative if hardware dependency is confirmed true alongside the existing life_safety_classification=true.
Secondary Path
The fallback option if the primary recommendation cannot proceed.
Secondary Path — RETAIN
If hardware_dependency is confirmed true during discovery, the structural pre-check fires (hardware_dependency=true AND life_safety_classification=true), and the recommendation must be revised to RETAIN. The ESI VM's role as a Linac integration point makes this a material risk that must be resolved before any migration planning proceeds.
Right-Sizing Notes
Why the cloud sizing estimate is set where it is, based on the telemetry available.
Fleet avg mem_active_pct = 18.0% across 3 powered-on VMs (msqapp: 14.0%, msqesi: 31.0%, msqsql: 9.0%). Under Rule T3, 18.0% < 25% would classify as SRC-RS-003 OVERSIZED. However, Rule T2 hard rule fires unconditionally: snapshot_count=1 AND apm_telemetry_available=false. A single point-in-time snapshot cannot distinguish a genuinely idle VM from one sampled between bursts; OVERSIZED and ZOMBIE classifications are FORBIDDEN. Tier is capped at SRC-RS-002 MODERATE (0.25 reduction). Rule T1 does not apply: application_type='Client/Server', not 'Hardware-Integrated'; life_safety_classification is absent from catalog. cpu_avg_pct is null for all 3 VMs — CPU utilization cannot be assessed (DEF-RS-CPU-NULL). All 3 VMs run Windows Server 2012 (EOL October 2023) with phi_handling=true, constituting active HIPAA Security Rule §164.312 compliance risk on all nodes. Telemetry depth gap surfaced as TELEMETRY-DEPTH-INSUFFICIENT.
This tab shows the raw output of each analysis agent — the specialized components that examine telemetry, dependencies, contracts, and cost. It is the most detailed and most technical view. Most readers will find the Overview and Backlog tabs sufficient.
Telemetry Agent
Utilization Profilemoderate
Snapshot Count1
APM AvailableFalse
Hard Rule TriggeredTrue
Right-Sizing Opp.True
Agent Confidence45%
Telemetry — VM Counts
Powered Off VMs0
Flagged VMs0
Mem Active % (avg)18.0%
Data WindowNone
Inventory Source
Dependency Agent
Epic IntegrationFalse
PHI HandledTrue
Hardware DependencyUNKNOWN
Latency SensitiveUNKNOWN
Blast Radiuscritical
Migration RiskHIGH
Conf (Inferred)72%
Conf (Confirmed)58%
Verificationsingle_source
Cloud Connectivity
ComplianceHIPAA / HITRUST
Upstream (2)
Radiation Oncology Linear Accelerator / Treatment Delivery Hardware (inferred): inferred: rvtools.vm_notes.msqesi — 'external system interface server RAD/ONC'; Mosaiq ESI is the standard Elekta…
Active Directory / Windows Authentication (assumed): inferred: application_type=Client/Server on Windows Server 2012; standard dependency pattern for Elekta Mosaiq…
Downstream (2)
Radiation Oncology Clinical Workflow / Treatment Planning (RAD/ONC): confirmed: rvtools.vm_notes.msqapp — 'Mosaiq 2.6 app server RAD/ONC'; rvtools.vm_notes.msqesi — 'external system…
Epic (downstream consumer — NEGATED): confirmed: catalog.epic_integration=false
Procurement Agent
SaaS Available—
SaaS VendorElekta
SaaS ProductElekta MOSAIQ Online
Repurchase ViableINSUFFICIENT_EVIDENCE
BAA ConfirmedUNKNOWN
SaaS Confidence
Inferred Score70%
Confirmed Score0%
Verificationsingle_source
AWS Equiv.None
Provisioning Agent (Python engine — not LLM)
Current VMs3
Recommended VMs2
Right-Sizing25% reduction
Current TCO/yr$19,500
Projected Cloud TCO$10,400
Migration One-Time$2,080
Dual-Running Cost$9,967
-$2,947
Year-1 Net
$9,100
Steady-State
$15,253
3-Year
AWS TargetAmazon EC2
Azure TargetAzure Virtual Machines
Cost Modeldefault_multipliers
DEF: {'default_id': 'DEF-WL-001', 'field': 'workload_type', 'value': 'app_server', 'reason': 'workload_type indeterminable'}
DEF: {'default_id': 'DEF-MIG-001', 'field': 'migration_factor', 'value': 0.2, 'reason': 'migration_complexity not provided'}
DEF: {'default_id': 'DEF-COST-001', 'field': 'cost_model_basis', 'value': 'default_multipliers', 'reason': 'no customer cloud quote provided; using SRC-TCO-* and SRC-MULT-* registry defaults'}
Siemens Syngo / Syngovia
Vendor: Siemens Healthineers · 12 VMs · Criticality: High · HIPAA / HITRUST — PHI handled; BAA requi
REPURCHASE
40%
Confidence
28%
Data Quality
ROI: Year-1 Net: -$19,760 · Steady-State: $31,200/yr · 3-Year: $42,640
Overview
Backlog
Provenance
Technical Details
Agents
Recommendation
REPURCHASE ⏳ PENDING CONDITIONS
SRC-CAP-001 Binding cap · Ceiling: 40% · Threshold: 60%
Secondary: RETAIN
Confidence
40%
Viable, but specific gaps must close first.
ROI Summary (Python-computed, not LLM)
-$19,760
Year-1 Net
$31,200
Steady-State /yr
$42,640
3-Year Total
ROI variance: ±60%
⚙ Financials computed by provisioning_engine_v3 (Python) — not LLM-generated
In plain terms: migrating costs money up front — Year 1 is a net outflow of $19,760. From Year 2 onward the app saves $31,200 every year. The 3-Year total nets these against each other.
Finding
Siemens Syngo / Syngovia is a strong REPURCHASE candidate — Siemens Healthineers offers Syngo.via Cloud and teamplay as direct SaaS successors, and the inferred SaaS confidence of 70 clears the threshold for a strategic recommendation. However, the BAA status with the SaaS vendor is unconfirmed and PHI is handled, which caps analytical confidence at 40 and places this recommendation in a 'pending conditions' state. Confirming the BAA, validating hardware integration dependencies (DICOM modalities, Citrix delivery layer), and obtaining a formal contract proposal from Siemens Healthineers are the three actions needed to unlock an executable migration plan.
In plain terms: this app should move to the vendor's cloud product, but a few specific items must be confirmed first. See Conditions to Unlock below.
Conditions to Unlock (2 items)
This recommendation becomes executable once these conditions are confirmed:
☐ Confirm Business Associate Agreement with Siemens Healthineers for Cloud Migration
Owner: Compliance Officer / Legal · Timeline: 10 days · Unlocks: compliance_close, cap_resolution
☐ Confirm Whether Syngo Connects to Physical Imaging Hardware (CT, MRI, Cath Lab)
Owner: Clinical Informatics Lead / Infrastructure Architect · Timeline: 20 days · Unlocks: migration_commit, operational_safety
See Backlog tab for full action backlog and P2/P3 items.
This is the list of work needed to move this recommendation forward. The 2 P1 items are the priority — they unlock the recommendation. P2 and P3 items improve confidence and reduce risk but are not blockers. 'Story points' measure effort, not time or cost — see the Methodology tab for how they work.
40%
Current
60%
Threshold
55%
Ceiling
24
Story Points
Why This Score
Confidence is hard-capped at 40 by SRC-CAP-001 (phi_handled=true, baa_confirmed=UNKNOWN) — the pre-cap score of 63.5 is irrelevant until the BAA is confirmed. Three additional caps (SRC-CAP-002 at 55, SRC-CAP-003 at 65, SRC-CAP-004 at 65) form a sequential stack that will bind in order once SRC-CAP-001 is lifted, meaning BAA confirmation alone will not reach the 60-point threshold without also resolving the majority-INFERRED dimension gap addressed by Sprint 2 and Sprint 3 items.
Proceeding with REPURCHASE planning before BAA confirmation and hardware dependency validation exposes the organization to HIPAA enforcement risk and the possibility that the recommendation reverses to RETAIN if physical DICOM hardware coupling is confirmed.
Action Backlog — 6 Items · 24 Story Points · Ceiling: 55%
APP-006-001 Confirm BAA with Siemens Healthineers for Syngo.via Cloud / teamplay SaaS
P1 BAA/Legal
Obtain and execute a Business Associate Agreement (BAA) with Siemens Healthineers covering Syngo.via Cloud and/or teamplay SaaS. phi_handled=true and baa_confirmed=UNKNOWN is the sole binding constraint (SRC-CAP-001) holding confidence at 40 — the lowest cap in the stack. No SaaS migration can proceed without a signed BAA. Compliance officer must retrieve any existing BAA with Siemens Healthineers
+15% confidence -0% ROI variance 👤 📋 · 3 pts
APP-006-002 Validate Hardware Dependency — DICOM Modality Integration and Physical Device Coupling
P1 Dependency Mapping
hardware_dependency=UNKNOWN is a CRITICAL gap that could block the REPURCHASE recommendation entirely. Siemens Syngo / Syngovia is a PACS and cardiology imaging platform that typically integrates directly with CT, MR, Echo, and Cath Lab modalities via DICOM. If hardware_dependency=true AND latency_sensitive=true, the structural pre-check fires and REPURCHASE is blocked in favor of RETAIN. Engage c
+0% confidence -0% ROI variance 👤 📋 · 5 pts
APP-006-003 Confirm Life Safety Classification for Syngo Dynamics Cardiology Workflows
P2 Security/Compliance
life_safety_classification=UNKNOWN. Syngo Dynamics handles cardiology studies and may support real-time cath lab or EP lab display workflows. If life_safety_classification=true alongside hardware_dependency=true, the structural RETAIN pre-check fires and REPURCHASE is blocked. Engage clinical informatics owners (Andy Little / Andy Ross per vm_notes) and the FLC POC contact (Mangual Rivera, Gilbert
+0% confidence -0% ROI variance 👤 📋 · 3 pts
APP-006-004 Obtain Siemens Healthineers Contract Intelligence and SaaS Migration Pricing
P2 Procurement
contract_intelligence=INSUFFICIENT_EVIDENCE. No contract ACV, renewal date, or SaaS migration pricing is on file. ROI variance is capped at 60% in part due to this gap. Procurement must: (1) retrieve the current Siemens Healthineers contract for Syngo / Syngovia including ACV, renewal date, and termination clauses; (2) request a customer-specific RFP response or migration feasibility assessment fo
+0% confidence -0% ROI variance 👤 📋 · 5 pts
APP-006-005 Deploy APM Telemetry and Collect 30-Day Multi-Snapshot Utilization Data
P3 Telemetry
apm_telemetry_available=false and snapshot_count=1. The current right-sizing recommendation (25% reduction, 12→9 VMs) is based on a single RVTools snapshot with 8.7% average memory utilization and null CPU data across all 12 VMs. SRC-RS-002 hard cap applies; OVERSIZED (SRC-RS-003) and ZOMBIE (SRC-RS-004) tiers are forbidden until multi-point telemetry is available. Deploy APM tooling (e.g., Azure
+0% confidence -0% ROI variance 👤 📋 · 5 pts
APP-006-006 Document DR and RTO/RPO Requirements for Multi-Facility Imaging Platform
P3 Architecture
dr_requirements=MISSING. No DR documentation exists for a High-criticality, multi-facility imaging platform spanning MDC-CITRIX and MUSC-USB datacenters serving Black River (FMC) and MUSC-USB clinical sites. This gap contributes to ROI variance and prevents cloud architecture finalization. Engage application owners (Andy Little, Andy Ross, Mike Slezak per vm_notes) and the business continuity team
+0% confidence -0% ROI variance 👤 📋 · 3 pts
This tab shows the evidence behind the recommendation. The seven dimensions are scored by quality — CONFIRMED evidence counts fully, INFERRED evidence counts less. The validator table confirms each analysis step passed its checks. Green means the step is sound.
Evidence Classification — 7 Dimensions
Dimension Classification Multiplier Contribution
Telemetry SINGLE 0.50 6.00
Dependency Mapping INFERRED 0.25 2.50
Procurement Contracts MISSING 0.00 0.00
Architecture Diagrams INFERRED 0.25 1.75
Security Compliance INFERRED 0.25 1.25
Business Criticality SINGLE 0.50 2.00
Dr Requirements MISSING 0.00 0.00
Applied Cap: SRC-CAP-001
Validation Reports — Per Agent
Agent Status Provenance Counts Defaults Used Rule Failures
telemetry PASS
SRC:2CUST:9COMP:4DEF:2
 None ✓ None
dependency PASS
SRC:12COMP:5DEF:5
 None ✓ None
procurement PASS
None ✓ None
provisioning PASS_WITH_DEFAULTS
SRC:7CUST:1COMP:9DEF:1
• {'default_id': 'DEF-WL-001', 'field': 'workload_type', 'value': 'app_server', 'r
• {'default_id': 'DEF-MIG-001', 'field': 'migration_factor', 'value': 0.2, 'reason
 ✓ None
synthesizer PASS
None ✓ None
confidence_advisor PASS
COMP:15
 None ✓ None
This section contains the full technical reasoning behind the recommendation: the rules that fired, the decision path, the confidence math, and the open intelligence gaps. It is intended for analysts and engineers. Executive readers can rely on the Overview tab.
Rules Applied
These are the specific governance rules that produced this recommendation.
▸ RULE 1 — REPURCHASE
A vendor SaaS replacement is available and credible, and the app has no hardware constraint — so the recommendation is to replace it with the vendor's cloud service.
▸ SRC-CAP-001 (binding)
Confidence is capped at 40 because the app handles PHI and the vendor's Business Associate Agreement is not confirmed. PHI cannot move to a cloud vendor without a confirmed BAA.
▸ SRC-CAP-002
Confidence is capped at 55 because more than half of the seven evidence dimensions are inferred or missing — the conclusion is data-thin.
▸ SRC-CAP-003
Confidence is capped at 65 because none of the seven evidence dimensions reached CONFIRMED quality — there is no high-quality anchor for the analysis.
▸ SRC-CAP-004
Confidence is capped at 65 because telemetry is a single point-in-time snapshot with no ongoing monitoring — right-sizing needs usage patterns over time.
▸ RULE T2 — Right-Sizing Hard Cap
Right-sizing is limited to a moderate 25% reduction because telemetry is a single snapshot. More aggressive right-sizing needs 30+ days of monitoring data.
Decision Path
The step-by-step trace of how the recommendation was reached, in the system's own terms.
PRE-CHECK: hardware_dependency=UNKNOWN, latency_sensitive=UNKNOWN, life_safety_classification=UNKNOWN — all UNKNOWN, pre-check does not fire. RULE 1 (REPURCHASE): saas_alternative_available=true (Syngo.via Cloud / teamplay per application notes and procurement agent), saas_confidence_inferred=70 >= 50, hardware_dependency=UNKNOWN != true — Rule 1 FIRES. strategic_recommendation=REPURCHASE. CONFIDENCE SCORING: base=50, telemetry(SINGLE)=+6.0, dependency_mapping(INFERRED)=+2.5, procurement_contracts(MISSING)=+0.0, architecture_diagrams(INFERRED)=+1.75, security_compliance(INFERRED)=+1.25, business_criticality(SINGLE)=+2.0, dr_requirements(MISSING)=+0.0, pre_cap_total=63.5. CAP EVALUATION: SRC-CAP-001 fires (phi_handled=true, baa_confirmed=UNKNOWN != true) → cap=40; SRC-CAP-002 fires (5 of 7 dimensions INFERRED/MISSING) → cap=55; SRC-CAP-003 fires (zero CONFIRMED) → cap=65; SRC-CAP-004 fires (snapshot_count=1, apm_telemetry_available=false) → cap=65. Lowest cap = SRC-CAP-001 at 40. final_confidence=40. GATE: final_confidence=40 < 60, strategic_recommendation=REPURCHASE (not RETAIN) → recommendation_status=PENDING_CONDITIONS, displayed_primary=REPURCHASE (v4.1: gate does not force RETAIN), confidence_gate_applied=true.
Cap Evaluation (all fired caps): SRC-CAP-001: PHI handled + BAA not confirmed → ceiling 40 (ceiling=40) ← BINDING SRC-CAP-002: ≥4 of 7 dimensions INFERRED/MISSING → ceiling 55 (ceiling=55) SRC-CAP-003: Zero CONFIRMED dimensions → ceiling 65 (ceiling=65) SRC-CAP-004: Single snapshot, no APM → ceiling 65 (ceiling=65)
Confidence Scoring Breakdown
How the confidence percentage was calculated, dimension by dimension.
Base score: 50 Evidence contributions: Telemetry SINGLE +6.00 Dependency Mapping INFERRED +2.50 Procurement Contracts MISSING +0.00 Architecture Diagrams INFERRED +1.75 Security Compliance INFERRED +1.25 Business Criticality SINGLE +2.00 Dr Requirements MISSING +0.00 Pre-cap score: 63.5 Binding cap applied: SRC-CAP-001 Final confidence: 40%
Intelligence Gaps
Specific things we don't yet know. Each gap is a reason the confidence score is held back.
BAA Status with Siemens Healthineers SaaS (Syngo.via Cloud / teamplay)
PHI is handled and BAA is unconfirmed. SRC-CAP-001 caps confidence at 40. No SaaS migration can proceed without a signed BAA. Obtain BAA documentation from Siemens Healthineers before any procurement decision.
Hardware Dependency Validation (DICOM Modality Integration)
hardware_dependency=UNKNOWN. Syngo / Syngovia is a PACS/cardiology imaging platform that typically integrates directly with CT, MR, Echo, and Cath Lab modalities via DICOM. If hardware_dependency=true AND latency_sensitive=true, the structural pre-check fires and REPURCHASE is blocked. Field investigation required before migration planning.
Life Safety Classification
life_safety_classification=UNKNOWN. Cardiology imaging (Syngo Dynamics) may support clinical decision-making in life-safety contexts. If confirmed true alongside hardware_dependency=true, structural RETAIN pre-check fires. Requires clinical stakeholder confirmation.
Contract Intelligence and Licensing Terms
contract_intelligence=INSUFFICIENT_EVIDENCE. Current contract terms, renewal dates, and SaaS migration pricing from Siemens Healthineers are unknown. ROI variance is capped at 60% due to this gap. Procurement must obtain current contract and SaaS pricing proposal.
Citrix Delivery Layer Disposition (MDC-CITRIX)
Two powered-off VMs (ctxsyngo00, ctxsyngovia64) in MDC-CITRIX datacenter indicate a legacy Citrix delivery path. It is unclear whether this is decommissioned or in standby. Disposition of Citrix infrastructure must be confirmed before migration scope is finalized.
APM Telemetry and Multi-Snapshot Utilization Data
apm_telemetry_available=false and snapshot_count=1. Right-sizing recommendation (25% reduction) is based on a single RVTools snapshot with 8.7% average memory utilization. Multi-point telemetry is required to validate right-sizing before cloud instance selection.
DR and RTO/RPO Requirements
dr_requirements=MISSING. No DR documentation available for a High-criticality, multi-facility imaging platform. DR requirements must be established before cloud architecture can be finalized.
Technical Detail
A fuller technical explanation of the recommendation and its constraints.
Procurement agent confirmed Siemens Healthineers offers Syngo.via Cloud and teamplay as direct SaaS successors with inferred confidence of 70, clearing the Rule 1 REPURCHASE threshold of 50. Dependency agent identified a high blast radius with confirmed multi-facility deployment (Black River FMC, MUSC-USB) and a Citrix delivery layer, but hardware_dependency remains UNKNOWN — a critical gap given Syngo's typical DICOM modality integration patterns. Telemetry agent flagged snapshot_count=1 with no APM telemetry, triggering SRC-RS-002 right-sizing (25% reduction, 12→9 VMs), and the fleet average memory utilization of 8.7% confirms moderate utilization. SRC-CAP-001 applies because phi_handled=true and baa_confirmed=UNKNOWN, capping confidence at 40 and placing the REPURCHASE recommendation in PENDING_CONDITIONS status pending BAA confirmation and hardware dependency validation.
Secondary Path
The fallback option if the primary recommendation cannot proceed.
Secondary Path — RETAIN
RETAIN is the secondary disposition if hardware dependency investigation confirms direct DICOM modality integration or life-safety classification that would trigger the structural pre-check blocker.
Right-Sizing Notes
Why the cloud sizing estimate is set where it is, based on the telemetry available.
Fleet avg mem active = 8.7% across 10 powered-on VMs with non-null data. Under Rule T3 this would qualify for SRC-RS-003 (OVERSIZED, <25%). However, Rule T2 fires unconditionally: snapshot_count=1 AND apm_telemetry_available=false. Hard rule caps maximum tier at SRC-RS-002 (MODERATE, 0.25 reduction). SRC-RS-003 and SRC-RS-004 are FORBIDDEN. Single-snapshot data cannot distinguish genuinely idle VMs from those sampled between burst cycles — this is the precise scenario T2 guards against. 12 VMs are powered off (ctxsyngo00, ctxsyngovia64) and contribute no utilization signal. CPU utilization data is absent for all 12 VMs. Telemetry depth is insufficient to support a higher-confidence right-sizing recommendation. Recommend collecting 30+ days of APM telemetry or a minimum of 2 RVTools snapshots before escalating tier.
This tab shows the raw output of each analysis agent — the specialized components that examine telemetry, dependencies, contracts, and cost. It is the most detailed and most technical view. Most readers will find the Overview and Backlog tabs sufficient.
Telemetry Agent
Utilization Profilemoderate
Snapshot Count1
APM AvailableFalse
Hard Rule TriggeredTrue
Right-Sizing Opp.True
Agent Confidence30%
Telemetry — VM Counts
Powered Off VMs2
Flagged VMs0
Mem Active % (avg)8.7%
Data WindowNone
Inventory Source
Dependency Agent
Epic IntegrationFalse
PHI HandledTrue
Hardware DependencyUNKNOWN
Latency SensitiveUNKNOWN
Blast Radiushigh
Migration RiskMEDIUM
Conf (Inferred)72%
Conf (Confirmed)48%
Verificationsingle_source
Cloud Connectivity
ComplianceHIPAA / HITRUST
Upstream (4)
DICOM Modality Sources (CT, MR, Echo, Cath Lab): inferred: Siemens Syngo product class (PACS/cardiology imaging platform receives DICOM from imaging modalities)
Hospital Information System / ADT Feed: inferred: standard Syngo Dynamics integration pattern for patient demographics and order routing
Active Directory / Identity Provider: inferred: Windows Server 2019/2022 domain-joined VMs in enterprise datacenter
vcockpit-infra (Syngo Steering Server): confirmed: rvtools.vm_notes.vcockpit-infra = 'Syngo Steering Server'
Downstream (5)
Radiology / Cardiology Workstations (clinical viewers): inferred: Syngo product class serves diagnostic review workstations; fmcsyngowks VM name pattern consistent with…
Black River Facility (FMC deployment): confirmed: rvtools.vm_notes on FMCSyngoApp1, FMCSyngoDB1, FMCSyngoEnc1, FMCSyngoMLA1 all reference 'Black River Syngo'…
Syngo Dynamics Cardiology Workflow (FLC POC): confirmed: rvtools.vm_notes.fmcsyngowks = 'FLC POC: Mangual Rivera, Gilberto Syngo Dynamics Test environment Mgr: Andy…
Syngovia Production Instance (syngoprod): confirmed: rvtools.vm_notes.syngoprod = 'Siemens Syngovia Requested by: Mike Slezak'
Citrix Delivery Layer (MDC-CITRIX datacenter): confirmed: ctxsyngo00 and ctxsyngovia64 are powered-off VMs in MDC-CITRIX datacenter with 'ctx' prefix indicating…
Procurement Agent
SaaS Available—
SaaS VendorSiemens Healthineers
SaaS ProductSyngo.via Cloud / teamplay
Repurchase ViableINSUFFICIENT_EVIDENCE
BAA ConfirmedUNKNOWN
SaaS Confidence
Inferred Score70%
Confirmed Score0%
Verificationsingle_source
AWS Equiv.None
Provisioning Agent (Python engine — not LLM)
Current VMs12
Recommended VMs9
Right-Sizing25% reduction
Current TCO/yr$78,000
Projected Cloud TCO$46,800
Migration One-Time$9,360
Dual-Running Cost$41,600
-$19,760
Year-1 Net
$31,200
Steady-State
$42,640
3-Year
AWS TargetAmazon EC2
Azure TargetAzure Virtual Machines
Cost Modeldefault_multipliers
DEF: {'default_id': 'DEF-WL-001', 'field': 'workload_type', 'value': 'app_server', 'reason': 'workload_type indeterminable'}
DEF: {'default_id': 'DEF-MIG-001', 'field': 'migration_factor', 'value': 0.2, 'reason': 'migration_complexity not provided'}
DEF: {'default_id': 'DEF-COST-001', 'field': 'cost_model_basis', 'value': 'default_multipliers', 'reason': 'no customer cloud quote provided; using SRC-TCO-* and SRC-MULT-* registry defaults'}
Sectra PACS
Vendor: Sectra · 124 VMs · Criticality: High · HIPAA / HITRUST — phi_handled=true; BAA
REPURCHASE
40%
Confidence
42%
Data Quality
ROI: Year-1 Net: -$204,187 · Steady-State: $322,400/yr · 3-Year: $440,613
Overview
Backlog
Provenance
Technical Details
Agents
Recommendation
REPURCHASE ⏳ PENDING CONDITIONS
SRC-CAP-001 Binding cap · Ceiling: 40% · Threshold: 60%
Secondary: REHOST
Confidence
40%
Viable, but specific gaps must close first.
ROI Summary (Python-computed, not LLM)
-$204,187
Year-1 Net
$322,400
Steady-State /yr
$440,613
3-Year Total
ROI variance: ±60%
⚙ Financials computed by provisioning_engine_v3 (Python) — not LLM-generated
In plain terms: migrating costs money up front — Year 1 is a net outflow of $204,187. From Year 2 onward the app saves $322,400 every year. The 3-Year total nets these against each other.
Finding
Sectra PACS is a strong candidate for migration to Sectra Medical Cloud — the same vendor offers a cloud-native SaaS version that would eliminate 124 on-premises VMs spread across seven MUSC campuses. However, before this migration can be formally recommended, MUSC must confirm that a Business Associate Agreement is in place with Sectra for the cloud platform, since this system handles protected health information. Once BAA status and contract terms are confirmed, this recommendation becomes executable with an estimated annual savings of approximately $322,000.
In plain terms: this app should move to the vendor's cloud product, but a few specific items must be confirmed first. See Conditions to Unlock below.
Conditions to Unlock (2 items)
This recommendation becomes executable once these conditions are confirmed:
☐ Verify Business Associate Agreement with Sectra for Cloud Hosting
Owner: Compliance Officer / Legal Counsel · Timeline: 10 days · Unlocks: compliance_close, cap_resolution
☐ Get Sectra Cloud Pricing and Contract Details
Owner: Procurement / Vendor Management · Timeline: 21 days · Unlocks: cost_clarity, migration_commit
See Backlog tab for full action backlog and P2/P3 items.
This is the list of work needed to move this recommendation forward. The 2 P1 items are the priority — they unlock the recommendation. P2 and P3 items improve confidence and reduce risk but are not blockers. 'Story points' measure effort, not time or cost — see the Methodology tab for how they work.
40%
Current
60%
Threshold
55%
Ceiling
27
Story Points
Why This Score
Sectra PACS carries a final confidence of 40 because SRC-CAP-001 is the binding cap: phi_handled=true is confirmed by catalog and RVTools DICOM evidence, while baa_confirmed=UNKNOWN because application_catalog.baa_valid is null and no BAA document is on file — under v3.1 tri-state rules, null cannot be inferred as true or false. Two additional caps (SRC-CAP-002 at 55 and SRC-CAP-004 at 65) are queued behind SRC-CAP-001 and will become binding in sequence as each prior cap is resolved through the actions in this backlog.
Proceeding with Sectra Medical Cloud migration without a confirmed BAA exposes MUSC to HIPAA Security Rule violations under 45 CFR §164.308(b) and §164.314(a), with potential OCR enforcement action and reputational harm across all 7 campus radiology operations.
Action Backlog — 6 Items · 27 Story Points · Ceiling: 55%
APP-007-001 Confirm BAA with Sectra for Medical Cloud Platform
P1 BAA/Legal
Retrieve or execute a Business Associate Agreement (BAA) between MUSC and Sectra covering the Sectra Medical Cloud SaaS platform. The catalog field baa_valid is null, which under v3.1 tri-state rules resolves to UNKNOWN. SRC-CAP-001 fires whenever phi_handled=true AND baa_confirmed != true, imposing a hard confidence ceiling of 40 and blocking migration execution. The required artifact is an execu
+15% confidence -0% ROI variance 👤 📋 · 3 pts
APP-007-002 Obtain Sectra Medical Cloud Commercial Proposal and Contract Terms
P1 Procurement
Request a formal commercial proposal from Sectra for Sectra Medical Cloud covering the full MUSC footprint: 124 VMs across 7 datacenters, 976 vCPUs, 2,904 GiB RAM, multi-site satellite architecture, DICOM routing nodes, AlwaysOn AG database clusters, 3D rendering, research environment, and business analytics sub-components. Current ROI projection of $322,400 annual savings carries a 60% variance b
+0% confidence -0% ROI variance 👤 📋 · 5 pts
APP-007-003 Assess Latency Requirements for Radiologist Reading Workstations
P2 Architecture
Determine whether PACS image retrieval for radiologist reading stations (6 UniView servers: sauniview01-06 across MUSC-USB and MUSC-ART) and the Sectra Wise clinical portal (SAWISEAPP01/02) has a quantified sub-second latency floor that would affect cloud viability. latency_sensitive is currently UNKNOWN because no vendor documentation or catalog field provides a confirmed latency threshold. If la
+0% confidence -0% ROI variance 👤 📋 · 3 pts
APP-007-004 Deploy 30-Day APM Telemetry Instrumentation Across 124-VM Fleet
P2 Telemetry
Instrument all 124 VMs with APM tooling (e.g., Datadog, Dynatrace, or native vRealize Operations) to collect 30+ days of CPU utilization, memory active percentage, DICOM throughput, storage I/O, and peak imaging load (CT/MRI/X-Ray ingestion rates). Current telemetry is a single point-in-time RVTools snapshot (snapshot_count=1, apm_telemetry_available=false), triggering SRC-CAP-004 (ceiling 65) and
+0% confidence -0% ROI variance 👤 📋 · 5 pts
APP-007-005 Map RIS and HL7 Upstream Integration Sources
P2 Dependency Mapping
Confirm the identity and integration method of the upstream Radiology Information System (RIS) or order entry system feeding DICOM Modality Worklist (DMWL) to Sectra PACS. catalog.epic_integration=false rules out a confirmed Epic HL7 feed but does not identify the actual worklist source. The upstream RIS/HL7 dependency is currently INFERRED from PACS architecture patterns (industry_pattern:PACS_RI
+0% confidence -0% ROI variance 👤 📋 · 3 pts
APP-007-006 Develop Multi-Site Cutover Sequencing and DR Validation Plan
P3 Architecture
Produce a validated cutover sequencing plan for migrating 124 VMs across 7 MUSC campus datacenters (MUSC-USB, MUSC-ART, MUSC-CHESTER, MUSC-FLORENCE, MUSC-LANCASTER, MUSC-MARION, MUSC-WAN) with zero downtime for active radiology operations. The plan must address: (1) AlwaysOn AG failover sequencing for SASHDB01/02 and SAWISEDB01/02 (20 vCPU/128 GiB synchronous pairs spanning USB↔ART); (2) satellite
+0% confidence -0% ROI variance 👤 📋 · 8 pts
This tab shows the evidence behind the recommendation. The seven dimensions are scored by quality — CONFIRMED evidence counts fully, INFERRED evidence counts less. The validator table confirms each analysis step passed its checks. Green means the step is sound.
Evidence Classification — 7 Dimensions
Dimension Classification Multiplier Contribution
Telemetry SINGLE 0.50 6.00
Dependency Mapping CONFIRMED 1.00 10.00
Procurement Contracts INFERRED 0.25 2.00
Architecture Diagrams INFERRED 0.25 1.75
Security Compliance INFERRED 0.25 1.25
Business Criticality CONFIRMED 1.00 4.00
Dr Requirements INFERRED 0.25 1.00
Applied Cap: SRC-CAP-001
Validation Reports — Per Agent
Agent Status Provenance Counts Defaults Used Rule Failures
telemetry PASS
SRC:2CUST:9COMP:5DEF:1
 None ✓ None
dependency PASS
SRC:12COMP:5DEF:5
 None ✓ None
procurement PASS
None ✓ None
provisioning PASS_WITH_DEFAULTS
SRC:7CUST:1COMP:9DEF:1
• {'default_id': 'DEF-WL-001', 'field': 'workload_type', 'value': 'app_server', 'r
• {'default_id': 'DEF-MIG-001', 'field': 'migration_factor', 'value': 0.2, 'reason
 ✓ None
synthesizer PASS
None ✓ None
confidence_advisor PASS
COMP:33
 None ✓ None
This section contains the full technical reasoning behind the recommendation: the rules that fired, the decision path, the confidence math, and the open intelligence gaps. It is intended for analysts and engineers. Executive readers can rely on the Overview tab.
Rules Applied
These are the specific governance rules that produced this recommendation.
▸ RULE 1 — REPURCHASE
A vendor SaaS replacement is available and credible — so the recommendation is to replace it with the vendor's cloud service.
▸ SRC-CAP-001 (binding)
Confidence is capped at 40 because the app handles PHI and the vendor's Business Associate Agreement is not confirmed. PHI cannot move to a cloud vendor without a confirmed BAA.
▸ SRC-CAP-002
Confidence is capped at 55 because more than half of the seven evidence dimensions are inferred or missing — the conclusion is data-thin.
▸ SRC-CAP-004
Confidence is capped at 65 because telemetry is a single point-in-time snapshot with no ongoing monitoring — right-sizing needs usage patterns over time.
▸ RULE T2 — Right-Sizing Hard Cap
Right-sizing is limited to a moderate 25% reduction because telemetry is a single snapshot. More aggressive right-sizing needs 30+ days of monitoring data.
Decision Path
The step-by-step trace of how the recommendation was reached, in the system's own terms.
PRE-CHECK: hardware_dependency=false → pre-check does not fire. RULE 1 (REPURCHASE): saas_alternative_available=true (catalog notes: 'SaaS alternative: Sectra Medical Cloud (SaaS)') AND saas_confidence_inferred=72 >= 50 AND hardware_dependency=false → Rule 1 fires. RULE 2 blocker check: decommission_annotation_present=false (0 flagged VMs, powered_off_vms=['dmrtr-6.5'] with no decommission annotation) → no blocker. STRATEGIC RECOMMENDATION = REPURCHASE. CONFIDENCE SCORING: base=50 + telemetry(SINGLE,12,0.50=6.0) + dependency(CONFIRMED,10,1.00=10.0) + procurement(INFERRED,8,0.25=2.0) + architecture(INFERRED,7,0.25=1.75) + security(INFERRED,5,0.25=1.25) + criticality(CONFIRMED,4,1.00=4.0) + dr(INFERRED,4,0.25=1.0) = 76.0 pre-cap. CAP EVALUATION: SRC-CAP-001 fires (phi_handled=true, baa_confirmed=UNKNOWN≠true) → cap=40; SRC-CAP-002 fires (5/7 dimensions INFERRED/SINGLE) → cap=55; SRC-CAP-004 fires (snapshot_count=1, apm_telemetry_available=false) → cap=65; SRC-CAP-003 does not fire (2 CONFIRMED dimensions). Lowest cap = SRC-CAP-001 = 40. FINAL CONFIDENCE = 40. GATE (v4.1): final_confidence=40 < 60 AND strategic_recommendation=REPURCHASE → recommendation_status=PENDING_CONDITIONS; displayed_primary=REPURCHASE (gate does NOT force RETAIN in v4.1). CONDITIONS TO UNLOCK: (1) Confirm BAA with Sectra for Medical Cloud platform; (2) Obtain Sectra Medical Cloud contract terms and pricing; (3) Assess latency requirements for radiologist reading stations; (4) Map RIS/HL7 upstream integration.
Cap Evaluation (all fired caps): SRC-CAP-001: PHI handled + BAA not confirmed → ceiling 40 (ceiling=40) ← BINDING SRC-CAP-002: ≥4 of 7 dimensions INFERRED/MISSING → ceiling 55 (ceiling=55) SRC-CAP-004: Single snapshot, no APM → ceiling 65 (ceiling=65)
Confidence Scoring Breakdown
How the confidence percentage was calculated, dimension by dimension.
Base score: 50 Evidence contributions: Telemetry SINGLE +6.00 Dependency Mapping CONFIRMED +10.00 Procurement Contracts INFERRED +2.00 Architecture Diagrams INFERRED +1.75 Security Compliance INFERRED +1.25 Business Criticality CONFIRMED +4.00 Dr Requirements INFERRED +1.00 Pre-cap score: 76.0 Binding cap applied: SRC-CAP-001 Final confidence: 40%
Intelligence Gaps
Specific things we don't yet know. Each gap is a reason the confidence score is held back.
BAA Confirmation — Sectra Medical Cloud
phi_handled=true and baa_confirmed=UNKNOWN; SRC-CAP-001 is the binding confidence cap at 40. No PHI-bearing SaaS migration can proceed without a confirmed BAA. Resolving this single gap raises the cap ceiling from 40 to 55 (next binding cap). Required artifact: executed BAA between MUSC and Sectra for Medical Cloud.
Sectra Medical Cloud Contract Terms and Pricing
saas_confidence_confirmed=0 and repurchase_viable=INSUFFICIENT_EVIDENCE. No contract intelligence available. ROI projection of $322,400 annual savings carries 60% variance until contract pricing is confirmed. Required artifact: Sectra Medical Cloud commercial proposal or existing contract.
Latency Sensitivity Assessment for Radiologist Workstations
latency_sensitive=UNKNOWN. PACS image retrieval for radiologist reading stations (6 UniView servers, SAWISEAPP01/02) may have sub-second latency requirements that affect cloud viability. If latency_sensitive=true is confirmed, migration complexity increases significantly and may require hybrid or edge-caching architecture.
APM / Application Performance Telemetry
apm_telemetry_available=false and snapshot_count=1. DICOM throughput, peak imaging load (CT/MRI/X-Ray ingestion rates), and storage I/O patterns are uncharacterized. Right-sizing reduction of 25% (SRC-RS-002) is a default assumption; actual savings may differ materially. Required: 30-day APM instrumentation or Sectra performance logs.
RIS / HL7 Upstream Integration Mapping
Upstream RIS/order entry system is inferred (not confirmed). catalog.epic_integration=false rules out confirmed Epic HL7 feed but does not identify the actual worklist source. Migration sequencing and integration testing scope cannot be finalized without this mapping.
Multi-site Cutover Sequencing Plan
7 campus satellite servers (Black River, Chester, Florence, Lancaster, Marion, Orangeburg, Sumter/SMP) require coordinated migration with no downtime for active radiology operations. No DR or cutover documentation available. Blast radius estimate is HIGH per dependency agent.
Technical Detail
A fuller technical explanation of the recommendation and its constraints.
Dependency agent confirms hardware_dependency=false and life_safety_classification=false, clearing the structural retain pre-check. Procurement agent identifies Sectra Medical Cloud as a same-vendor SaaS alternative with saas_confidence_inferred=72, satisfying Rule 1's threshold of 50. The 124-VM footprint across 7 MUSC datacenters (MUSC-USB, MUSC-ART, MUSC-CHESTER, MUSC-FLORENCE, MUSC-LANCASTER, MUSC-MARION, MUSC-WAN) represents a compelling consolidation target. SRC-CAP-001 fires as the binding cap (phi_handled=true, baa_confirmed=UNKNOWN), reducing final confidence to 40 and triggering PENDING_CONDITIONS status — the strategic recommendation remains REPURCHASE but requires BAA confirmation and contract validation before execution.
Secondary Path
The fallback option if the primary recommendation cannot proceed.
Secondary Path — REHOST
REHOST is the secondary disposition: all 124 VMs run on Windows Server 2019 (with CentOS 9 DICOM/satellite nodes), hardware_dependency=false, and no confirmed latency constraint — a standard IaaS lift-and-shift to AWS/Azure is technically viable as a fallback if Sectra Medical Cloud contract terms are unfavorable or latency requirements preclude SaaS.
Right-Sizing Notes
Why the cloud sizing estimate is set where it is, based on the telemetry available.
Fleet avg mem active pct = 14.7% across 124 VMs (single point-in-time snapshot). Under RULE T3, this would classify as SRC-RS-003 OVERSIZED (fleet avg < 25%). However, RULE T2 fires: snapshot_count=1 AND apm_telemetry_available=false. Tier is hard-capped at SRC-RS-002 MODERATE (0.25 reduction). Single-snapshot data cannot distinguish genuinely idle VMs from those sampled between bursts; this reasoning is the refusal trigger per v3.1. CPU utilization data is unavailable for all 124 VMs (cpu_avg_pct absent from all RVTools records). No memory pressure detected across fleet. 1 VM (dmrtr-6.5) is powered off. 1 VM (dmrtr-6.5) runs CentOS 7 (EOL June 30, 2024) and is flagged HIPAA-EOL-OTHER given phi_handled=true. TELEMETRY-DEPTH-INSUFFICIENT flag raised fleet-wide; 30+ days APM telemetry recommended before finalizing right-sizing.
This tab shows the raw output of each analysis agent — the specialized components that examine telemetry, dependencies, contracts, and cost. It is the most detailed and most technical view. Most readers will find the Overview and Backlog tabs sufficient.
Telemetry Agent
Utilization Profilemoderate
Snapshot Count1
APM AvailableFalse
Hard Rule TriggeredTrue
Right-Sizing Opp.True
Agent Confidence55%
Telemetry — VM Counts
Powered Off VMs1
Flagged VMs0
Mem Active % (avg)14.7%
Data Window1 days
Inventory Source
Dependency Agent
Epic IntegrationFalse
PHI HandledTrue
Hardware DependencyFalse
Latency SensitiveUNKNOWN
Blast Radiushigh
Migration RiskMEDIUM
Conf (Inferred)82%
Conf (Confirmed)71%
Verificationconfirmed
Cloud Connectivity
ComplianceHIPAA / HITRUST
Upstream (4)
Modality Systems / DICOM Sources (CT, MRI, X-Ray, etc.): inferred: DICOM SCP servers (sadtdm01 'Sectra DICOM SCP Server', sadtdm02 'Sectra Downtime Image Server') imply…
RIS / Order Entry System (likely Epic or standalone RIS): inferred: PACS systems universally receive HL7 orders/worklist from RIS or EHR. catalog.epic_integration=false rules…
Active Directory / Identity Provider: inferred: Windows Server 2019 domain-joined VMs across 7 datacenters imply AD dependency for authentication. Not…
Network Infrastructure (WAN / MUSC-WAN datacenter): confirmed: rvtools.datacenters includes MUSC-WAN with 30+ VMs routed through WAN fabric. Satellite servers (SABMCSAT01,…
Downstream (9)
Radiologist Workstations / Sectra UniView Viewers: confirmed: rvtools.vm.sauniview01-06.note = 'Sectra UniView USB/ART Windows Server 2019'. 6 UniView servers across…
Sectra Wise Clinical Portal (downstream image viewing): confirmed: rvtools.vm.SAWISEAPP01/02 ('USB Windows Server 2019'), SAWISEDB01 ('Sectra WISEDB AoAG Node 1'), SAWISEDB02…
Sectra ELS (Enterprise Linking Service / Load Balancing): confirmed: rvtools.vm.SAELS01-06 and SADTELS01-03 across MUSC-USB and MUSC-ART. ELS servers are downstream…
Sectra Research Platform (SARSH*): confirmed: rvtools.vm.SARSHDB01 ('Sectra Research Database Server 1'), SARSHDB02 ('Sectra Research Database Server 2'),…
Sectra Business Analytics / Data Warehouse (SASDWHDB01): confirmed: rvtools.vm.SASDWHDB01.note = 'Sectra Business Analytics database (SDWHDB) Team = Radiology Contact = Wendy…
Procurement Agent
SaaS Available—
SaaS VendorSectra
SaaS ProductSectra Medical Cloud
Repurchase ViableINSUFFICIENT_EVIDENCE
BAA ConfirmedUNKNOWN
SaaS Confidence
Inferred Score72%
Confirmed Score0%
Verificationsingle_source
AWS Equiv.None
Provisioning Agent (Python engine — not LLM)
Current VMs124
Recommended VMs93
Right-Sizing25% reduction
Current TCO/yr$806,000
Projected Cloud TCO$483,600
Migration One-Time$96,720
Dual-Running Cost$429,867
-$204,187
Year-1 Net
$322,400
Steady-State
$440,613
3-Year
AWS TargetAmazon EC2
Azure TargetAzure Virtual Machines
Cost Modeldefault_multipliers
DEF: {'default_id': 'DEF-WL-001', 'field': 'workload_type', 'value': 'app_server', 'reason': 'workload_type indeterminable'}
DEF: {'default_id': 'DEF-MIG-001', 'field': 'migration_factor', 'value': 0.2, 'reason': 'migration_complexity not provided'}
DEF: {'default_id': 'DEF-COST-001', 'field': 'cost_model_basis', 'value': 'default_multipliers', 'reason': 'no customer cloud quote provided; using SRC-TCO-* and SRC-MULT-* registry defaults'}
Philips IntelliSpace
Vendor: Philips · 66 VMs · Criticality: High · HIPAA / HITRUST — BAA with Philips Intel
REPURCHASE
55%
Confidence
28%
Data Quality
ROI: Year-1 Net: -$112,667 · Steady-State: $169,000/yr · 3-Year: $225,333
Overview
Backlog
Provenance
Technical Details
Agents
Recommendation
REPURCHASE ⏳ PENDING CONDITIONS
SRC-CAP-002 Binding cap · Ceiling: 55% · Threshold: 60%
Secondary: REHOST
Confidence
55%
Viable, but specific gaps must close first.
ROI Summary (Python-computed, not LLM)
-$112,667
Year-1 Net
$169,000
Steady-State /yr
$225,333
3-Year Total
ROI variance: ±55%
⚙ Financials computed by provisioning_engine_v3 (Python) — not LLM-generated
In plain terms: migrating costs money up front — Year 1 is a net outflow of $112,667. From Year 2 onward the app saves $169,000 every year. The 3-Year total nets these against each other.
Finding
Philips IntelliSpace is a strong REPURCHASE candidate: the same vendor (Philips) offers a cloud-hosted cardiovascular informatics platform (IntelliSpace Cloud) that would eliminate 66 on-premises VMs, 52 of which are currently powered off. The transition cannot be executed today because BAA status, contract terms, and hardware integration details are unconfirmed — a focused 4-week discovery sprint covering these three items would unlock the migration path. Once conditions are met, the move to Philips IntelliSpace Cloud is expected to reduce annual infrastructure costs by approximately $169,000.
In plain terms: this app should move to the vendor's cloud product, but a few specific items must be confirmed first. See Conditions to Unlock below.
Conditions to Unlock (3 items)
This recommendation becomes executable once these conditions are confirmed:
☐ Verify Business Associate Agreement with Philips for Cloud Platform
Owner: Legal / Compliance Officer · Timeline: 10 days · Unlocks: compliance_close, cap_resolution, migration_commit
☐ Confirm Whether Cardiac Imaging Hardware Requires On-Premises Connectivity
Owner: Enterprise Architect / Cardiology IT Lead · Timeline: 20 days · Unlocks: migration_commit, cap_resolution, operational_safety
☐ Confirm That IntelliSpace Processes Patient Health Information
Owner: Privacy Officer / Application Owner · Timeline: 10 days · Unlocks: compliance_close, cap_resolution
See Backlog tab for full action backlog and P2/P3 items.
This is the list of work needed to move this recommendation forward. The 3 P1 items are the priority — they unlock the recommendation. P2 and P3 items improve confidence and reduce risk but are not blockers. 'Story points' measure effort, not time or cost — see the Methodology tab for how they work.
55%
Current
60%
Threshold
62%
Ceiling
24
Story Points
Why This Score
The current confidence of 55 is bound by SRC-CAP-002 (5 of 7 evidence dimensions are INFERRED or MISSING: dependency_mapping, procurement_contracts, architecture_diagrams, security_compliance, and dr_requirements), which imposes a hard ceiling of 55 regardless of the pre-cap calculated score of 61.75. Because current_confidence equals the ceiling, no backlog item can produce a confidence delta in this run — the backlog is structured to resolve the binding cap by upgrading INFERRED/MISSING dimensions to CONFIRMED, which will lift the SRC-CAP-002 ceiling and allow confidence to climb above 55 in a subsequent scoring cycle.
Executing REPURCHASE without resolving the three CRITICAL gaps (BAA status, hardware dependency, PHI confirmation) risks initiating a SaaS migration that is either legally non-compliant under HIPAA or technically infeasible due to unconfirmed cardiac imaging hardware integration requirements.
Action Backlog — 6 Items · 24 Story Points · Ceiling: 62%
APP-008-001 Confirm BAA Status with Philips for IntelliSpace Cloud
P1 BAA/Legal
Engage Philips account team and internal Legal/Compliance to determine whether a signed BAA exists or can be executed for Philips IntelliSpace Cloud. Retrieve or initiate BAA documentation. A signed BAA is a legal prerequisite before any PHI can be processed in the SaaS environment and is required to resolve baa_confirmed=UNKNOWN, which is the primary blocker preventing SRC-CAP-001 from being eval
+0% confidence -0% ROI variance 👤 📋 · 3 pts
APP-008-002 Verify Hardware Dependency for Cardiac Imaging Modalities
P1 Dependency Mapping
Conduct a structured interview with the Cardiology IT team and Philips technical account manager to determine whether IntelliSpace Cardiovascular (ISCV) requires proprietary hardware interfaces (DICOM gateway, HL7 broker, dedicated acquisition workstations) that cannot be served from a cloud-hosted SaaS platform. Document all upstream modality connections (echo machines, stress test equipment, OBG
+0% confidence -0% ROI variance 👤 📋 · 5 pts
APP-008-003 Formally Confirm PHI Handling Classification for IntelliSpace
P1 Security/Compliance
Obtain explicit written confirmation from the application owner or Privacy Officer that Philips IntelliSpace processes Protected Health Information (PHI), resolving the field-name mismatch between catalog.phi_handling=true and the governed phi_handled boolean. This confirmation is required to formally trigger SRC-CAP-001 evaluation and to scope BAA requirements accurately. The dependency agent ass
+7% confidence -0% ROI variance 👤 📋 · 3 pts
APP-008-004 Retrieve Philips On-Premises Contract Terms and Renewal Date
P2 Procurement
Engage Procurement and the Philips account team to retrieve the current on-premises IntelliSpace contract, including ACV, renewal date, termination clauses, and any migration credits or cloud transition incentives. procurement_contracts=MISSING is contributing a 15-point penalty to ROI variance and preventing confirmation of transition timing. Additionally, request a Philips IntelliSpace Cloud quo
+7% confidence -0% ROI variance 👤 📋 · 5 pts
APP-008-005 Deploy 30-Day APM Telemetry Collection Across Active VM Fleet
P2 Telemetry
Deploy APM agents or enable extended RVTools/vRealize collection across the 14 powered-on VMs (6 MUSC-WAN server-class + 8 MDC-CITRIX thin-client endpoints) to capture CPU, memory, disk I/O, and network utilization over a minimum 30-day window. Current telemetry is a single point-in-time RVTools snapshot with 9.1% VM data coverage, triggering SRC-CAP-004 and the RULE T2 hard cap that limits right-
+0% confidence -0% ROI variance 👤 📋 · 5 pts
APP-008-006 Remediate Windows Server 2012 End-of-Support Instances
P3 Security/Compliance
RMC-ISP-PROD and RMC-ISP-TEST are running Windows Server 2012 (64-bit), which reached end of extended support in October 2023. These VMs carry phi_handling=true and represent an active HIPAA Security Rule violation under 45 CFR 164.312 (Technical Safeguards) — flagged as HIPAA-EOL-002 by the telemetry agent. Remediation is required regardless of migration path: (a) if REPURCHASE proceeds, these VM
+7% confidence -0% ROI variance 👤 📋 · 3 pts
This tab shows the evidence behind the recommendation. The seven dimensions are scored by quality — CONFIRMED evidence counts fully, INFERRED evidence counts less. The validator table confirms each analysis step passed its checks. Green means the step is sound.
Evidence Classification — 7 Dimensions
Dimension Classification Multiplier Contribution
Telemetry SINGLE 0.50 6.00
Dependency Mapping INFERRED 0.25 2.50
Procurement Contracts MISSING 0.00 0.00
Architecture Diagrams MISSING 0.00 0.00
Security Compliance INFERRED 0.25 1.25
Business Criticality SINGLE 0.50 2.00
Dr Requirements MISSING 0.00 0.00
Applied Cap: SRC-CAP-002
Validation Reports — Per Agent
Agent Status Provenance Counts Defaults Used Rule Failures
telemetry PASS
SRC:2CUST:9COMP:3DEF:2
 None ✓ None
dependency PASS
SRC:4CUST:3COMP:5DEF:4
 None ✓ None
procurement PASS
None ✓ None
provisioning PASS_WITH_DEFAULTS
SRC:7CUST:1COMP:9DEF:1
• {'default_id': 'DEF-WL-001', 'field': 'workload_type', 'value': 'app_server', 'r
• {'default_id': 'DEF-MIG-001', 'field': 'migration_factor', 'value': 0.2, 'reason
 ✓ None
synthesizer PASS
None ✓ None
confidence_advisor PASS
COMP:15
 None ✓ None
This section contains the full technical reasoning behind the recommendation: the rules that fired, the decision path, the confidence math, and the open intelligence gaps. It is intended for analysts and engineers. Executive readers can rely on the Overview tab.
Rules Applied
These are the specific governance rules that produced this recommendation.
▸ RULE 1 — REPURCHASE
A vendor SaaS replacement is available and credible, and the app has no hardware constraint — so the recommendation is to replace it with the vendor's cloud service.
▸ SRC-CAP-002 (binding)
Confidence is capped at 55 because more than half of the seven evidence dimensions are inferred or missing — the conclusion is data-thin.
▸ SRC-CAP-004
Confidence is capped at 65 because telemetry is a single point-in-time snapshot with no ongoing monitoring — right-sizing needs usage patterns over time.
▸ RULE T2 — Right-Sizing Hard Cap
Right-sizing is limited to a moderate 25% reduction because telemetry is a single snapshot. More aggressive right-sizing needs 30+ days of monitoring data.
Decision Path
The step-by-step trace of how the recommendation was reached, in the system's own terms.
PRE-CHECK: hardware_dependency=UNKNOWN, latency_sensitive=UNKNOWN, life_safety_classification=UNKNOWN — all tri-state UNKNOWN, pre-check does not fire. | RULE 1 (REPURCHASE): saas_alternative_available=true (SOURCED: catalog.notes — 'SaaS alternative: Philips IntelliSpace Cloud'), saas_confidence_inferred=68 ≥ 50, hardware_dependency=UNKNOWN ≠ true — ALL conditions met. RULE 1 FIRES. strategic_recommendation=REPURCHASE. | CONFIDENCE SCORING: base=50, telemetry=SINGLE(+6.0), dependency_mapping=INFERRED(+2.5), procurement_contracts=MISSING(+0), architecture_diagrams=MISSING(+0), security_compliance=INFERRED(+1.25), business_criticality=SINGLE(+2.0), dr_requirements=MISSING(+0). Pre-cap total=61.75. | CAP EVALUATION: SRC-CAP-001 — phi_handled=UNKNOWN, not confirmed true, does not fire. SRC-CAP-002 — 5 of 7 dimensions INFERRED/MISSING (dependency_mapping, procurement_contracts, architecture_diagrams, security_compliance, dr_requirements), majority threshold met, FIRES at 55. SRC-CAP-003 — zero CONFIRMED dimensions, FIRES at 65. SRC-CAP-004 — snapshot_count=1 AND apm_telemetry_available=false, FIRES at 65. Lowest cap = SRC-CAP-002 at 55. final_confidence=55. | GATE (v4.1): final_confidence=55 < 60, strategic_recommendation=REPURCHASE (not RETAIN) → recommendation_status=PENDING_CONDITIONS, confidence_gate_applied=true, displayed_primary=REPURCHASE (gate does NOT force RETAIN under v4.1). | ROI VARIANCE: base=5 + telemetry_penalty=0(SINGLE) + procurement_penalty=15(MISSING) + arch_penalty=15(MISSING) + dep_penalty=10(INFERRED) + dr_penalty=10(MISSING) = 55%, below 60% cap.
Cap Evaluation (all fired caps): SRC-CAP-002: ≥4 of 7 dimensions INFERRED/MISSING → ceiling 55 (ceiling=55) ← BINDING SRC-CAP-004: Single snapshot, no APM → ceiling 65 (ceiling=65)
Confidence Scoring Breakdown
How the confidence percentage was calculated, dimension by dimension.
Base score: 50 Evidence contributions: Telemetry SINGLE +6.00 Dependency Mapping INFERRED +2.50 Procurement Contracts MISSING +0.00 Architecture Diagrams MISSING +0.00 Security Compliance INFERRED +1.25 Business Criticality SINGLE +2.00 Dr Requirements MISSING +0.00 Pre-cap score: 61.75 Binding cap applied: SRC-CAP-002 Final confidence: 55%
Intelligence Gaps
Specific things we don't yet know. Each gap is a reason the confidence score is held back.
BAA Status with Philips IntelliSpace Cloud
baa_confirmed=UNKNOWN blocks HIPAA-compliant SaaS transition; a signed BAA with Philips for the cloud platform is a legal prerequisite before any PHI can be processed in the SaaS environment. Must be resolved before REPURCHASE can be executed.
Hardware Dependency Verification
hardware_dependency=UNKNOWN; Philips IntelliSpace cardiovascular informatics platforms often integrate with dedicated cardiac imaging hardware (echo machines, stress test equipment) via proprietary DICOM or HL7 interfaces. If confirmed hardware dependency exists, the pre-check structural blocker may apply and change the recommendation to RETAIN.
PHI Handling Confirmation
phi_handled=UNKNOWN despite cardiovascular informatics context strongly implying PHI (echo reports, stress test results, OBG imaging). Confirmation is required to determine SRC-CAP-001 applicability and to scope BAA requirements.
Procurement Contract Intelligence
contract_intelligence=INSUFFICIENT_EVIDENCE; existing Philips on-premises contract terms (renewal date, termination clauses, migration credits) are unknown. Contract discovery is required to determine transition timing and cost.
52 Powered-Off VM Disposition
52 of 66 VMs (MDC-CITRIX) are powered off with no decommission annotation. It is unclear whether these are decommissioned Citrix endpoints, standby capacity, or orphaned VMs. Formal disposition is required before migration scope can be finalized.
Windows Server 2012 End-of-Support
RMC-ISP-PROD and RMC-ISP-TEST are running Windows Server 2012 (64-bit), which reached end of extended support in October 2023. These VMs represent an active security and compliance liability under HIPAA/HITRUST and must be addressed regardless of migration path.
APM Telemetry Unavailable
apm_telemetry_available=false with snapshot_count=1; application-layer performance baselines are absent, making it impossible to validate SaaS performance equivalence or establish migration success criteria without additional instrumentation.
Citrix Delivery Layer Migration Path
60 of 66 VMs reside in MDC-CITRIX datacenter; if Philips IntelliSpace Cloud delivers a browser-based or native client, the Citrix delivery layer may be eliminated — but this dependency must be confirmed with Philips cloud platform architecture documentation.
Technical Detail
A fuller technical explanation of the recommendation and its constraints.
Rule 1 (REPURCHASE) fires: catalog notes confirm a SaaS alternative exists (Philips IntelliSpace Cloud, same vendor), procurement agent reports saas_confidence_inferred=68 ≥ 50, and hardware_dependency=UNKNOWN (not confirmed true), satisfying all three Rule 1 conditions. The pre-check structural blocker does not apply because hardware_dependency, latency_sensitive, and life_safety_classification are all UNKNOWN rather than confirmed true. Telemetry data reveals 52 of 66 VMs are powered off with fleet average memory utilization of 9.8%, strongly supporting a right-sizing and SaaS transition case. Confidence is capped at 55 by SRC-CAP-002 (5 of 7 dimensions INFERRED or MISSING) and SRC-CAP-004 (snapshot_count=1, APM unavailable), placing the recommendation in PENDING_CONDITIONS status pending BAA confirmation, contract discovery, and hardware dependency verification.
Secondary Path
The fallback option if the primary recommendation cannot proceed.
Secondary Path — REHOST
REHOST is the secondary disposition if SaaS transition is blocked by hardware integration constraints discovered during dependency verification — the active MUSC-WAN server VMs (RMC-ISCV-CORE, RMC-ISCV-ANALYTICS, RMC-ISCV-WEB) are standard Windows Server workloads with no confirmed hardware tie-in and could be lifted to IaaS while the 52 powered-off Citrix endpoint VMs are decommissioned.
Right-Sizing Notes
Why the cloud sizing estimate is set where it is, based on the telemetry available.
RULE T2 hard cap applied: snapshot_count=1 AND apm_telemetry_available=false. Maximum allowable tier is SRC-RS-002 (MODERATE, 0.25 reduction); SRC-RS-003 (OVERSIZED) and SRC-RS-004 (ZOMBIE) are FORBIDDEN. Raw fleet_avg_mem_active_pct of 9.8% (derived from only 6 of 66 VMs, 9.1% data coverage) would satisfy the OVERSIZED threshold under RULE T3, but that reasoning is explicitly named as a refusal trigger under v3.1 T2. 52 of 66 VMs are powered off with null mem_active_pct; 8 additional powered-on VMs in MDC-CITRIX also have null mem_active_pct. CPU utilization data is absent for all 66 VMs. Right-sizing opportunity is flagged at MODERATE tier pending 30-day APM telemetry collection. RULE T1 does not apply: application_type is Client/Server, not Hardware-Integrated.
This tab shows the raw output of each analysis agent — the specialized components that examine telemetry, dependencies, contracts, and cost. It is the most detailed and most technical view. Most readers will find the Overview and Backlog tabs sufficient.
Telemetry Agent
Utilization Profilemoderate
Snapshot Count1
APM AvailableFalse
Hard Rule TriggeredTrue
Right-Sizing Opp.True
Agent Confidence28%
Telemetry — VM Counts
Powered Off VMs52
Flagged VMs0
Mem Active % (avg)9.8%
Data Window1 days
Inventory Source
Dependency Agent
Epic IntegrationFalse
PHI HandledUNKNOWN
Hardware DependencyUNKNOWN
Latency SensitiveUNKNOWN
Blast Radiushigh
Migration RiskMEDIUM
Conf (Inferred)72%
Conf (Confirmed)38%
Verificationsingle_source
Cloud Connectivity
ComplianceHIPAA / HITRUST
Upstream (4)
Cardiac Imaging Modalities (Echo / Stress): inferred: rvtools.vm_notes RMC-ISCV-ANALYTICS, RMC-ISCV-CORE — 'Echo and stress exams' implies upstream DICOM feed from…
OBG Imaging Modalities: inferred: rvtools.vm_prefix — obgiscv01-08 VMs suggest upstream OBG imaging or clinical data feed
Active Directory / Identity Provider: inferred: industry_pattern — Windows Server 2016/2012 client-server deployment in healthcare enterprise typically…
Citrix Delivery Infrastructure (MDC-CITRIX): confirmed: rvtools.datacenter — 52 powered-off and 8 powered-on VMs reside in MDC-CITRIX datacenter, indicating…
Downstream (3)
Cardiology Clinicians / Reading Workstations: inferred: industry_pattern — ISCV is a cardiovascular informatics platform; downstream consumers are cardiologists and…
IntelliSpace Portal (RMC-ISP-PROD / RMC-ISP-TEST): confirmed: rvtools.vm_notes — RMC-ISP-PROD and RMC-ISP-TEST notes read 'Philips IntelliSpace Portal', indicating a…
EHR / Clinical Data Repository: inferred: industry_pattern — cardiovascular informatics platforms in healthcare typically push structured results to…
Procurement Agent
SaaS Available—
SaaS VendorPhilips
SaaS ProductPhilips IntelliSpace Cloud (cloud-hosted cardiovascular informatics)
Repurchase ViableINSUFFICIENT_EVIDENCE
BAA ConfirmedUNKNOWN
SaaS Confidence
Inferred Score68%
Confirmed Score0%
Verificationsingle_source
AWS Equiv.None
Provisioning Agent (Python engine — not LLM)
Current VMs66
Recommended VMs50
Right-Sizing25% reduction
Current TCO/yr$429,000
Projected Cloud TCO$260,000
Migration One-Time$52,000
Dual-Running Cost$229,667
-$112,667
Year-1 Net
$169,000
Steady-State
$225,333
3-Year
AWS TargetAmazon EC2
Azure TargetAzure Virtual Machines
Cost Modeldefault_multipliers
DEF: {'default_id': 'DEF-WL-001', 'field': 'workload_type', 'value': 'app_server', 'reason': 'workload_type indeterminable'}
DEF: {'default_id': 'DEF-MIG-001', 'field': 'migration_factor', 'value': 0.2, 'reason': 'migration_complexity not provided'}
DEF: {'default_id': 'DEF-COST-001', 'field': 'cost_model_basis', 'value': 'default_multipliers', 'reason': 'no customer cloud quote provided; using SRC-TCO-* and SRC-MULT-* registry defaults'}
MUSE
Vendor: GE Healthcare · 8 VMs · Criticality: High · HIPAA / HITRUST — phi_handled=true; life
REPURCHASE
40%
Confidence
28%
Data Quality
ROI: Year-1 Net: -$13,173 · Steady-State: $20,800/yr · 3-Year: $28,427
Overview
Backlog
Provenance
Technical Details
Agents
Recommendation
REPURCHASE ⏳ PENDING CONDITIONS
SRC-CAP-001 Binding cap · Ceiling: 40% · Threshold: 60%
Secondary: REHOST
Confidence
40%
Viable, but specific gaps must close first.
ROI Summary (Python-computed, not LLM)
-$13,173
Year-1 Net
$20,800
Steady-State /yr
$28,427
3-Year Total
ROI variance: ±50%
⚙ Financials computed by provisioning_engine_v3 (Python) — not LLM-generated
In plain terms: migrating costs money up front — Year 1 is a net outflow of $13,173. From Year 2 onward the app saves $20,800 every year. The 3-Year total nets these against each other.
Finding
GE Healthcare offers a cloud-hosted version of MUSE, making a vendor SaaS transition the preferred path to eliminate 8 on-premises VMs and reduce operational overhead. However, the BAA status with GE Healthcare's cloud offering is unconfirmed and no contract intelligence is available, which prevents a high-confidence recommendation at this time. A targeted discovery sprint to validate the BAA, confirm SaaS feature parity for cardiology workflows, and document clinical continuity requirements will unlock this recommendation.
In plain terms: this app should move to the vendor's cloud product, but a few specific items must be confirmed first. See Conditions to Unlock below.
Conditions to Unlock (2 items)
This recommendation becomes executable once these conditions are confirmed:
☐ Verify Business Associate Agreement with GE HealthCare — Required Before Any Migration Can Proceed
Owner: Compliance Officer / Legal · Timeline: 21 days · Unlocks: compliance_close, cap_resolution
☐ Get Formal Pricing and Migration Proposal from GE HealthCare for Cloud-Hosted MUSE
Owner: Procurement / Vendor Management · Timeline: 21 days · Unlocks: cost_clarity, migration_commit
See Backlog tab for full action backlog and P2/P3 items.
This is the list of work needed to move this recommendation forward. The 2 P1 items are the priority — they unlock the recommendation. P2 and P3 items improve confidence and reduce risk but are not blockers. 'Story points' measure effort, not time or cost — see the Methodology tab for how they work.
40%
Current
60%
Threshold
55%
Ceiling
36
Story Points
Why This Score
Confidence is hard-capped at 40 by SRC-CAP-001 (phi_handled=true AND baa_confirmed=UNKNOWN), which is the binding and lowest of four simultaneously active caps (SRC-CAP-001 at 40, SRC-CAP-002 at 55, SRC-CAP-003 at 65, SRC-CAP-004 at 65); the pre-cap score of 63.5 is irrelevant until the BAA is confirmed. No confidence improvement is achievable through any backlog action until SRC-CAP-001 is resolved — BAA confirmation (APP-009-001) is the sole gate to unlocking headroom.
Proceeding with MUSE migration planning without BAA confirmation creates active HIPAA liability under 45 CFR §164.314 for a life-safety classified cardiology system handling ECG waveforms and cardiac monitoring data for a high-criticality production environment.
Action Backlog — 6 Items · 36 Story Points · Ceiling: 55%
APP-009-001 Confirm BAA with GE HealthCare Technologies Inc. (SRC-CAP-001 Resolution)
P1 BAA/Legal
Obtain and review a current, executed Business Associate Agreement (BAA) with GE HealthCare Technologies Inc. (the post-2023 spin-off legal entity). The existing catalog field baa_valid is null, triggering SRC-CAP-001 and capping confidence at 40. Compliance officer must confirm: (1) whether a BAA exists with the current GE HealthCare legal entity; (2) whether any legacy BAA with 'GE Healthcare' o
+15% confidence -0% ROI variance 👤 📋 · 5 pts
APP-009-002 Engage GE HealthCare Account Team for SaaS Contract Intelligence and Migration Proposal
P1 Procurement
Engage the GE HealthCare account team to obtain: (1) a formal SaaS/cloud-hosted MUSE NX migration proposal scoped to MUSC's deployment (8 VMs, dual-datacenter, HL7 interface environment); (2) ACV and renewal date for the existing on-premises license; (3) SaaS pricing tiers and contract terms; (4) migration support clauses and implementation timeline. Current contract_intelligence=INSUFFICIENT_EVID
+0% confidence -0% ROI variance 👤 📋 · 5 pts
APP-009-003 Validate GE HealthCare SaaS Feature Parity and HL7 Interface Compatibility for MUSC Cardiology Workflows
P2 Vendor Engagement
Conduct a structured feature parity assessment between the current on-premises MUSE NX deployment and GE HealthCare's cloud-hosted offering. Scope must include: (1) HL7 message type compatibility (ADT, ORU, ORM) with existing ECG acquisition devices (MAC series, CASE systems) in a cloud-hosted topology; (2) downstream EMR integration continuity — catalog confirms epic_integration=false but dedicat
+0% confidence -0% ROI variance 👤 📋 · 8 pts
APP-009-004 Develop and Validate Life-Safety Clinical Continuity Plan for MUSE Migration Cutover
P2 Security/Compliance
Develop a validated clinical continuity plan covering cardiology workflow downtime during MUSE migration cutover. life_safety_classification=true and blast_radius_estimate=critical; dr_requirements=MISSING. Required deliverables: (1) documented RTO/RPO targets for the MUSE cardiology information system aligned with MUSC clinical operations; (2) downtime procedures for ECG acquisition, waveform rev
+0% confidence -0% ROI variance 👤 📋 · 8 pts
APP-009-005 Deploy APM Telemetry and Collect 30-Day Performance Baseline for MUSE Fleet
P2 Telemetry
Deploy application performance monitoring (APM) across the 7 active MUSE NX production and test VMs to collect a minimum 30-day time-series baseline. apm_telemetry_available=false and snapshot_count=1, triggering SRC-CAP-004 and the Rule T2 hard cap (right-sizing capped at MODERATE/SRC-RS-002). CPU utilization is entirely absent (cpu_avg_pct null on all 8 VMs). Required data: (1) CPU and RAM utili
+0% confidence -0% ROI variance 👤 📋 · 5 pts
APP-009-006 Confirm HL7 Interface Inventory and Downstream EMR Integration Dependencies
P3 Dependency Mapping
Produce a confirmed interface inventory for the MUSE NX deployment, replacing the current INFERRED dependency classifications. Scope: (1) enumerate all upstream ECG acquisition devices and cardiology modalities sending data to MUSE via HL7 or GE MAC protocols, including device model, interface type, message types (ADT, ORU), and volumes; (2) identify the downstream EMR system receiving cardiology
+0% confidence -0% ROI variance 👤 📋 · 5 pts
This tab shows the evidence behind the recommendation. The seven dimensions are scored by quality — CONFIRMED evidence counts fully, INFERRED evidence counts less. The validator table confirms each analysis step passed its checks. Green means the step is sound.
Evidence Classification — 7 Dimensions
Dimension Classification Multiplier Contribution
Telemetry SINGLE 0.50 6.00
Dependency Mapping INFERRED 0.25 2.50
Procurement Contracts MISSING 0.00 0.00
Architecture Diagrams INFERRED 0.25 1.75
Security Compliance INFERRED 0.25 1.25
Business Criticality SINGLE 0.50 2.00
Dr Requirements MISSING 0.00 0.00
Applied Cap: SRC-CAP-001
Validation Reports — Per Agent
Agent Status Provenance Counts Defaults Used Rule Failures
telemetry PASS
SRC:2CUST:9COMP:2DEF:2
 None ✓ None
dependency PASS
SRC:7CUST:3COMP:3DEF:2
 None ✓ None
procurement PASS
None ✓ None
provisioning PASS_WITH_DEFAULTS
SRC:7CUST:1COMP:9DEF:1
• {'default_id': 'DEF-WL-001', 'field': 'workload_type', 'value': 'app_server', 'r
• {'default_id': 'DEF-MIG-001', 'field': 'migration_factor', 'value': 0.2, 'reason
 ✓ None
synthesizer PASS
None ✓ None
confidence_advisor PASS
COMP:15
 None ✓ None
This section contains the full technical reasoning behind the recommendation: the rules that fired, the decision path, the confidence math, and the open intelligence gaps. It is intended for analysts and engineers. Executive readers can rely on the Overview tab.
Rules Applied
These are the specific governance rules that produced this recommendation.
▸ RULE 1 — REPURCHASE
A vendor SaaS replacement is available and credible, and the app has no hardware constraint — so the recommendation is to replace it with the vendor's cloud service.
▸ SRC-CAP-001 (binding)
Confidence is capped at 40 because the app handles PHI and the vendor's Business Associate Agreement is not confirmed. PHI cannot move to a cloud vendor without a confirmed BAA.
▸ SRC-CAP-002
Confidence is capped at 55 because more than half of the seven evidence dimensions are inferred or missing — the conclusion is data-thin.
▸ SRC-CAP-003
Confidence is capped at 65 because none of the seven evidence dimensions reached CONFIRMED quality — there is no high-quality anchor for the analysis.
▸ SRC-CAP-004
Confidence is capped at 65 because telemetry is a single point-in-time snapshot with no ongoing monitoring — right-sizing needs usage patterns over time.
▸ RULE T2 — Right-Sizing Hard Cap
Right-sizing is limited to a moderate 25% reduction because telemetry is a single snapshot. More aggressive right-sizing needs 30+ days of monitoring data.
Decision Path
The step-by-step trace of how the recommendation was reached, in the system's own terms.
PRE-CHECK: hardware_dependency=false — structural blocker does not fire. RULE 2 (RETIRE): No decommission annotation on production application; one powered-off dev VM (retired-2024-10-09-musedev) is a dev environment artifact, not an application-level decommission annotation — Rule 2 does not fire. RULE 1 (REPURCHASE): saas_alternative_available=true (procurement), saas_confidence_inferred=68 >= 50, hardware_dependency=false — all three conditions met. RULE 1 FIRES. strategic_recommendation=REPURCHASE. CONFIDENCE GATE (v4.1): Base=50, telemetry(SINGLE)=+6.0, dependency_mapping(INFERRED)=+2.5, procurement_contracts(MISSING)=+0.0, architecture_diagrams(INFERRED)=+1.75, security_compliance(INFERRED)=+1.25, business_criticality(SINGLE)=+2.0, dr_requirements(MISSING)=+0.0. Pre-cap=63.5. Cap evaluation: SRC-CAP-001 (phi_handled=true, baa_confirmed=UNKNOWN)=40; SRC-CAP-002 (5/7 dimensions INFERRED/MISSING)=55; SRC-CAP-003 (zero CONFIRMED)=65; SRC-CAP-004 (snapshot_count=1, apm=false)=65. Lowest cap=SRC-CAP-001=40. final_confidence=40. Gate: 40 < 60 AND strategic_recommendation=REPURCHASE (not RETAIN) → displayed_primary=REPURCHASE, recommendation_status=PENDING_CONDITIONS, confidence_gate_applied=true. v4.1 gate does NOT force displayed_primary to RETAIN.
Cap Evaluation (all fired caps): SRC-CAP-001: PHI handled + BAA not confirmed → ceiling 40 (ceiling=40) ← BINDING SRC-CAP-002: ≥4 of 7 dimensions INFERRED/MISSING → ceiling 55 (ceiling=55) SRC-CAP-003: Zero CONFIRMED dimensions → ceiling 65 (ceiling=65) SRC-CAP-004: Single snapshot, no APM → ceiling 65 (ceiling=65)
Confidence Scoring Breakdown
How the confidence percentage was calculated, dimension by dimension.
Base score: 50 Evidence contributions: Telemetry SINGLE +6.00 Dependency Mapping INFERRED +2.50 Procurement Contracts MISSING +0.00 Architecture Diagrams INFERRED +1.75 Security Compliance INFERRED +1.25 Business Criticality SINGLE +2.00 Dr Requirements MISSING +0.00 Pre-cap score: 63.5 Binding cap applied: SRC-CAP-001 Final confidence: 40%
Intelligence Gaps
Specific things we don't yet know. Each gap is a reason the confidence score is held back.
BAA Confirmation with GE Healthcare SaaS
SRC-CAP-001 binding cap at 40. phi_handled=true requires a confirmed BAA with GE Healthcare's cloud hosting entity before any data migration. Without BAA confirmation, REPURCHASE cannot proceed to EXECUTABLE status. Must obtain and review GE Healthcare cloud BAA terms.
Procurement Contract Intelligence
contract_intelligence=INSUFFICIENT_EVIDENCE and saas_confidence_confirmed=0. No contract terms, renewal dates, SaaS pricing, or migration support clauses are available. Required to validate SaaS transition cost model and timeline.
SaaS Feature Parity and Clinical Workflow Validation
saas_confidence_inferred=68 but confirmed=0. GE Healthcare's cloud MUSE offering must be validated for feature parity with current on-premises deployment, particularly HL7 interface compatibility with existing ECG acquisition devices and downstream EMR integration.
Life-Safety Clinical Continuity Plan
life_safety_classification=true with blast_radius_estimate=critical. No DR documentation available (dr_requirements=MISSING). A validated clinical continuity plan covering cardiology workflow downtime during migration cutover is required before any transition can be approved.
APM Telemetry and Application Performance Baseline
apm_telemetry_available=false. Actual application response times, transaction volumes, and peak load patterns are unknown. Required to size GE Healthcare SaaS tier appropriately and validate SLA commitments for cardiology workflows.
HL7 Interface Dependency Mapping
Upstream ECG acquisition devices and downstream EMR integration are inferred from VM naming and vendor product knowledge. Confirmed interface inventory (HL7 message types, endpoints, volumes) is required to plan interface migration to SaaS environment.
DR Requirements and RTO/RPO Documentation
dr_requirements=MISSING. Dual-DC presence (MUSC-ART, MUSC-USB) implies DR intent but no confirmed RTO/RPO targets exist. Required to evaluate whether GE Healthcare SaaS meets MUSC's cardiology system availability requirements.
EOL VM Disposition
retired-2024-10-09-musedev is powered off and running Windows Server 2008 R2 (EOL). Confirm this VM is formally decommissioned and remove from inventory to eliminate EOL OS compliance exposure.
Technical Detail
A fuller technical explanation of the recommendation and its constraints.
The procurement agent identifies GE Healthcare's cloud/SaaS-hosted MUSE deployment as available with inferred confidence of 68, clearing the Rule 1 REPURCHASE threshold (>=50) and with no hardware dependency to trigger the structural pre-check. The dependency agent confirms hardware_dependency=false and latency_sensitive=false, removing the two structural blockers for cloud migration despite life_safety_classification=true. The telemetry agent flags a right-sizing opportunity (SRC-RS-002, 25% reduction viable) and avg memory active of 17.3%, supporting the case that the current 8-VM footprint is oversized for actual workload. However, SRC-CAP-001 applies because phi_handled=true and baa_confirmed=UNKNOWN, capping confidence at 40 and placing the recommendation in PENDING_CONDITIONS status pending BAA validation and procurement contract confirmation.
Secondary Path
The fallback option if the primary recommendation cannot proceed.
Secondary Path — REHOST
REHOST is the secondary disposition if SaaS transition is not viable — the 7 active production VMs on Windows Server 2019 are lift-and-shift compatible with no hardware constraints, and the dual-DC architecture maps naturally to a multi-AZ cloud deployment pattern.
Right-Sizing Notes
Why the cloud sizing estimate is set where it is, based on the telemetry available.
Fleet avg mem active pct is 17.3%, which under Rule T3 thresholds would qualify for SRC-RS-003 (OVERSIZED, <25%). However, Rule T2 (single-snapshot hard rule) is unconditionally triggered: snapshot_count=1 and apm_telemetry_available=false. Maximum allowed tier is capped at SRC-RS-002 (MODERATE, 0.25 reduction). SRC-RS-003 (OVERSIZED) and SRC-RS-004 (ZOMBIE) are forbidden. A single point-in-time snapshot cannot safely distinguish a genuinely idle VM from one sampled between burst cycles — this is precisely the unsafe inference Rule T2 exists to prevent. Additional telemetry (30+ days APM time-series or multiple RVTools snapshots) is required before OVERSIZED or ZOMBIE classification can be safely applied. CPU utilization data is entirely absent (cpu_avg_pct null on all 8 VMs), further limiting confidence. One VM (retired-2024-10-09-musedev) is powered off, carries a retirement date in its name, and runs Windows Server 2008 R2 — an EOL OS since January 2020 with active HIPAA compliance implications given phi_handling=true. That VM must be formally decommissioned or confirmed data-sanitized and removed from inventory.
This tab shows the raw output of each analysis agent — the specialized components that examine telemetry, dependencies, contracts, and cost. It is the most detailed and most technical view. Most readers will find the Overview and Backlog tabs sufficient.
Telemetry Agent
Utilization Profilemoderate
Snapshot Count1
APM AvailableFalse
Hard Rule TriggeredTrue
Right-Sizing Opp.True
Agent Confidence45%
Telemetry — VM Counts
Powered Off VMs1
Flagged VMs0
Mem Active % (avg)17.3%
Data WindowNone
Inventory Source
Dependency Agent
Epic IntegrationFalse
PHI HandledTrue
Hardware DependencyFalse
Latency SensitiveFalse
Blast Radiuscritical
Migration RiskHIGH
Conf (Inferred)78%
Conf (Confirmed)52%
Verificationsingle_source
Cloud Connectivity
ComplianceHIPAA / HITRUST
Upstream (3)
ECG Acquisition Devices / Cardiology Modalities: inferred: GE MUSE is a cardiology information system that receives waveform data from ECG carts, stress test systems,…
HL7 Interface Engine / Integration Layer: inferred: vm_note on MUSEPRODNXINT='Clark Houle. MUSE Production.' and MUSETESTNXINT='MUSE test interface server. Clark…
Active Directory / Windows Authentication: inferred: all production VMs run Windows Server 2019; GE MUSE Client/Server architecture requires domain authentication
Downstream (3)
Epic EHR (or equivalent EMR): inferred: catalog.epic_integration=false (SOURCED) rules out direct Epic integration; however, MUSE commonly feeds…
Cardiology Reporting / PACS: inferred: GE MUSE commonly integrates with cardiovascular PACS or reporting systems for waveform archival and physician…
Clinical Workstations (Web Client): inferred: MUSEPRODNXWEB vm_note='Clark Houle MUSE Production server' — dedicated web VM implies browser-based clinical…
Procurement Agent
SaaS Available—
SaaS VendorGE Healthcare (GE HealthCare Technologies Inc.)
SaaS ProductMUSE Cardiology Information System — Cloud / SaaS-hosted deployment
Repurchase ViableINSUFFICIENT_EVIDENCE
BAA ConfirmedUNKNOWN
SaaS Confidence
Inferred Score68%
Confirmed Score0%
Verificationinferred_only
AWS Equiv.None
Provisioning Agent (Python engine — not LLM)
Current VMs8
Recommended VMs6
Right-Sizing25% reduction
Current TCO/yr$52,000
Projected Cloud TCO$31,200
Migration One-Time$6,240
Dual-Running Cost$27,733
-$13,173
Year-1 Net
$20,800
Steady-State
$28,427
3-Year
AWS TargetAmazon EC2
Azure TargetAzure Virtual Machines
Cost Modeldefault_multipliers
DEF: {'default_id': 'DEF-WL-001', 'field': 'workload_type', 'value': 'app_server', 'reason': 'workload_type indeterminable'}
DEF: {'default_id': 'DEF-MIG-001', 'field': 'migration_factor', 'value': 0.2, 'reason': 'migration_complexity not provided'}
DEF: {'default_id': 'DEF-COST-001', 'field': 'cost_model_basis', 'value': 'default_multipliers', 'reason': 'no customer cloud quote provided; using SRC-TCO-* and SRC-MULT-* registry defaults'}
Lumedx Apollo / CardioPACS
Vendor: Lumedx · 3 VMs · Criticality: High · HIPAA / HITRUST — PHI handled; BAA requi
REPURCHASE
40%
Confidence
32%
Data Quality
ROI: Year-1 Net: -$2,947 · Steady-State: $9,100/yr · 3-Year: $15,253
Overview
Backlog
Provenance
Technical Details
Agents
Recommendation
REPURCHASE ⏳ PENDING CONDITIONS
SRC-CAP-001 Binding cap · Ceiling: 40% · Threshold: 60%
Secondary: REHOST
Confidence
40%
Viable, but specific gaps must close first.
ROI Summary (Python-computed, not LLM)
-$2,947
Year-1 Net
$9,100
Steady-State /yr
$15,253
3-Year Total
ROI variance: ±50%
⚙ Financials computed by provisioning_engine_v3 (Python) — not LLM-generated
In plain terms: migrating costs money up front — Year 1 is a net outflow of $2,947. From Year 2 onward the app saves $9,100 every year. The 3-Year total nets these against each other.
Finding
Lumedx Apollo / CardioPACS is a strong candidate for migration to Lumedx's own cloud-hosted SaaS offering, which would eliminate three aging Windows Server 2012 virtual machines and reduce on-premises cardiovascular imaging infrastructure. However, the recommendation cannot be executed today: the vendor BAA has not been confirmed, SaaS pricing and contract terms are unverified, and an active data migration from the legacy Xcelera platform is still in progress. Completing BAA confirmation and a SaaS contract review are the two actions that will unlock this recommendation.
In plain terms: this app should move to the vendor's cloud product, but a few specific items must be confirmed first. See Conditions to Unlock below.
Conditions to Unlock (3 items)
This recommendation becomes executable once these conditions are confirmed:
☐ Confirm Business Associate Agreement with Lumedx for cloud migration
Owner: Legal / Compliance Officer · Timeline: 10 days · Unlocks: compliance_close, cap_resolution
☐ Get Lumedx cloud pricing and contract details to validate the SaaS migration path
Owner: Procurement / Vendor Management · Timeline: 21 days · Unlocks: migration_commit, cost_clarity, cap_resolution
☐ Confirm whether CardioPACS has physical hardware connections that would block cloud migration
Owner: Clinical Informatics / Infrastructure Architect · Timeline: 10 days · Unlocks: migration_commit, operational_safety
See Backlog tab for full action backlog and P2/P3 items.
This is the list of work needed to move this recommendation forward. The 3 P1 items are the priority — they unlock the recommendation. P2 and P3 items improve confidence and reduce risk but are not blockers. 'Story points' measure effort, not time or cost — see the Methodology tab for how they work.
40%
Current
60%
Threshold
55%
Ceiling
21
Story Points
Why This Score
Confidence is hard-capped at 40 by SRC-CAP-001 because phi_handled=true and baa_confirmed=UNKNOWN (catalog.baa_valid=null); the pre-cap score of 63.5 is irrelevant until the BAA artifact is obtained and the cap is lifted. No backlog item can move the needle above 40 in this run — all confidence_delta values are zero per B2 ceiling-bound rule, and the backlog is oriented entirely toward cap resolution and ROI variance reduction.
Proceeding with any cloud or SaaS migration for this PHI-bearing cardiovascular PACS without a confirmed BAA and verified hardware dependency status creates material HIPAA compliance exposure and risks triggering the structural retain blocker that would override the REPURCHASE recommendation.
Action Backlog — 6 Items · 21 Story Points · Ceiling: 55%
APP-010-001 Obtain signed BAA with Lumedx covering cloud-hosted SaaS deployment
P1 BAA/Legal
Retrieve or execute a signed Business Associate Agreement (BAA) with Lumedx that explicitly covers the cloud-hosted Apollo / CardioPACS SaaS deployment. The catalog field baa_valid is null, leaving baa_confirmed=UNKNOWN. SRC-CAP-001 caps confidence at 40 because phi_handled=true and baa_confirmed!=true. Resolving this to baa_confirmed=true is the single action that lifts the binding cap and allows
+15% confidence -0% ROI variance 👤 📋 · 3 pts
APP-010-002 Engage Lumedx procurement for SaaS pricing, contract terms, and customer reference
P1 Procurement
Initiate a formal procurement engagement with Lumedx to obtain: (1) cloud-hosted Apollo / CV Suite SaaS pricing for the cardiovascular service line profile (Ped Echo, Cath, EP), (2) contract terms including SLA, data residency, and HIPAA addendum, (3) at least one peer-institution reference for a comparable SaaS deployment. Currently saas_confidence_confirmed=0 and repurchase_viable=INSUFFICIENT_E
+0% confidence -0% ROI variance 👤 📋 · 5 pts
APP-010-003 Verify hardware dependency — DICOM modality and acquisition interface inventory
P1 Dependency Mapping
Conduct a field interview with Cath Lab and EP Lab staff and review Lumedx vendor documentation to determine whether Apollo / CardioPACS has physical hardware controller dependencies: hemodynamic recorders, EP mapping system interfaces (e.g., CARTO, EnSite), fluoroscopy DICOM gateways, or proprietary modality worklist servers requiring local network adjacency. hardware_dependency=UNKNOWN; if confi
+0% confidence -0% ROI variance 👤 📋 · 3 pts
APP-010-004 Confirm life-safety classification with clinical informatics and risk management
P2 Security/Compliance
Engage clinical informatics leadership and the institutional risk management office to formally classify whether Lumedx Apollo / CardioPACS meets the life-safety system threshold under institutional policy. The dependency agent assessed this as 'likely_life_safety' given Cath/EP procedural context, but life_safety_classification=UNKNOWN per v3.1 tri-state rules. Confirmation of true would: (1) ele
+0% confidence -0% ROI variance 👤 📋 · 2 pts
APP-010-005 Document DR/RTO requirements for High-criticality cardiovascular imaging system
P2 Architecture
Engage the application owner and clinical operations team to document RTO and RPO requirements for Lumedx Apollo / CardioPACS. dr_requirements=MISSING, contributing a 10-point penalty to ROI variance. For a High-criticality PHI-bearing cardiovascular PACS serving Cath/EP workflows, RTO is likely sub-4-hour; this must be confirmed and used to evaluate SaaS SLA adequacy and cloud architecture design
+0% confidence -0% ROI variance 👤 📋 · 3 pts
APP-010-006 Deploy continuous APM telemetry and confirm Xcelera migration completion timeline
P2 Telemetry
Two parallel actions: (1) Deploy APM tooling (e.g., Datadog, New Relic, or Azure Monitor agent) on all three VMs to collect multi-day CPU, memory, DICOM I/O, and network throughput baselines. Single-snapshot RVTools data (snapshot_count=1, apm_telemetry_available=false) caps right-sizing at SRC-RS-002 (MODERATE, 25% reduction) and prevents OVERSIZED/ZOMBIE classification. Multi-day APM data would
+0% confidence -0% ROI variance 👤 📋 · 5 pts
This tab shows the evidence behind the recommendation. The seven dimensions are scored by quality — CONFIRMED evidence counts fully, INFERRED evidence counts less. The validator table confirms each analysis step passed its checks. Green means the step is sound.
Evidence Classification — 7 Dimensions
Dimension Classification Multiplier Contribution
Telemetry SINGLE 0.50 6.00
Dependency Mapping INFERRED 0.25 2.50
Procurement Contracts MISSING 0.00 0.00
Architecture Diagrams INFERRED 0.25 1.75
Security Compliance INFERRED 0.25 1.25
Business Criticality SINGLE 0.50 2.00
Dr Requirements MISSING 0.00 0.00
Applied Cap: SRC-CAP-001
Validation Reports — Per Agent
Agent Status Provenance Counts Defaults Used Rule Failures
telemetry PASS
SRC:2CUST:14COMP:3DEF:2
 None ✓ None
dependency PASS
SRC:7CUST:4COMP:3DEF:3
 None ✓ None
procurement PASS
None ✓ None
provisioning PASS_WITH_DEFAULTS
SRC:7CUST:1COMP:9DEF:1
• {'default_id': 'DEF-WL-001', 'field': 'workload_type', 'value': 'app_server', 'r
• {'default_id': 'DEF-MIG-001', 'field': 'migration_factor', 'value': 0.2, 'reason
 ✓ None
synthesizer PASS
None ✓ None
confidence_advisor PASS
COMP:15
 None ✓ None
This section contains the full technical reasoning behind the recommendation: the rules that fired, the decision path, the confidence math, and the open intelligence gaps. It is intended for analysts and engineers. Executive readers can rely on the Overview tab.
Rules Applied
These are the specific governance rules that produced this recommendation.
▸ RULE 1 — REPURCHASE
A vendor SaaS replacement is available and credible, and the app has no hardware constraint — so the recommendation is to replace it with the vendor's cloud service.
▸ SRC-CAP-001 (binding)
Confidence is capped at 40 because the app handles PHI and the vendor's Business Associate Agreement is not confirmed. PHI cannot move to a cloud vendor without a confirmed BAA.
▸ SRC-CAP-002
Confidence is capped at 55 because more than half of the seven evidence dimensions are inferred or missing — the conclusion is data-thin.
▸ RULE T2 — Right-Sizing Hard Cap
Right-sizing is limited to a moderate 25% reduction because telemetry is a single snapshot. More aggressive right-sizing needs 30+ days of monitoring data.
Decision Path
The step-by-step trace of how the recommendation was reached, in the system's own terms.
PRE-CHECK: hardware_dependency=UNKNOWN, latency_sensitive=UNKNOWN, life_safety_classification=UNKNOWN — none confirmed true; pre-check does not fire. RULE 1 (REPURCHASE): saas_alternative_available=true, saas_confidence_inferred=60 ≥ 50, hardware_dependency=UNKNOWN ≠ true — all conditions met; RULE 1 FIRES. strategic_recommendation=REPURCHASE. CONFIDENCE SCORING: base=50, telemetry(SINGLE)=+6.0, dependency(INFERRED)=+2.5, procurement(MISSING)=+0.0, architecture(INFERRED)=+1.75, security(INFERRED)=+1.25, criticality(SINGLE)=+2.0, dr(MISSING)=+0.0; pre-cap=63.5. CAP EVALUATION: SRC-CAP-001 (phi=true, baa≠true)→40; SRC-CAP-002 (5/7 INFERRED/MISSING)→55; SRC-CAP-003 (zero CONFIRMED)→65; SRC-CAP-004 (snapshot=1, no APM)→65. Binding cap: SRC-CAP-001→40. GATE (v4.1): final_confidence=40 < 60, strategic_recommendation=REPURCHASE → displayed_primary=REPURCHASE, recommendation_status=PENDING_CONDITIONS, confidence_gate_applied=true. ROI VARIANCE: base=5 + telemetry(SINGLE→0) + procurement(MISSING→+15) + arch(INFERRED→+10) + dep(INFERRED→+10) + dr(MISSING→+10) = 50%; cap not reached.
Cap Evaluation (all fired caps): SRC-CAP-001: PHI handled + BAA not confirmed → ceiling 40 (ceiling=40) ← BINDING SRC-CAP-002: ≥4 of 7 dimensions INFERRED/MISSING → ceiling 55 (ceiling=55)
Confidence Scoring Breakdown
How the confidence percentage was calculated, dimension by dimension.
Base score: 50 Evidence contributions: Telemetry SINGLE +6.00 Dependency Mapping INFERRED +2.50 Procurement Contracts MISSING +0.00 Architecture Diagrams INFERRED +1.75 Security Compliance INFERRED +1.25 Business Criticality SINGLE +2.00 Dr Requirements MISSING +0.00 Pre-cap score: 63.5 Binding cap applied: SRC-CAP-001 Final confidence: 40%
Intelligence Gaps
Specific things we don't yet know. Each gap is a reason the confidence score is held back.
BAA confirmation absent — PHI handling confirmed
phi_handled=true but baa_confirmed=UNKNOWN; SRC-CAP-001 caps confidence at 40 and blocks EXECUTABLE status. Must obtain signed BAA with Lumedx (SaaS) or current on-prem vendor before any cloud migration can proceed.
Hardware dependency unverified — DICOM modality integration unknown
CardioPACS receives DICOM studies from Echo, Cath, and EP lab modalities; hardware_dependency=UNKNOWN. If physical DICOM gateway hardware or proprietary acquisition interfaces are present, the pre-check structural retain blocker may fire and override REPURCHASE. Field verification required.
SaaS contract and pricing unconfirmed — saas_confidence_confirmed=0
Lumedx cloud-hosted offering is inferred from vendor product catalog only; no contract terms, pricing, or migration scope have been validated. repurchase_viable=INSUFFICIENT_EVIDENCE. Procurement engagement with Lumedx required to confirm SaaS path viability.
Active Xcelera-to-Lumedx data migration in progress
lxxlinterfacep VM note confirms an active image and data migration from Xcelera (legacy cardiovascular imaging platform) to Lumedx production. Migration cutover to SaaS must be sequenced after this transition completes; premature migration risks data loss or workflow disruption.
DR and RTO requirements undocumented
No DR data available for a High-criticality cardiovascular imaging system. RTO/RPO requirements are unknown; this gap affects cloud architecture design and SaaS SLA negotiation.
Technical Detail
A fuller technical explanation of the recommendation and its constraints.
Rule 1 (REPURCHASE) fires: Lumedx offers a cloud-hosted SaaS product (Lumedx Apollo Cloud / CV Suite Cloud-Hosted), saas_confidence_inferred=60 meets the ≥50 threshold, and hardware_dependency=UNKNOWN does not equal true, so the hardware blocker does not apply. The pre-check structural retain blocker does not fire because hardware_dependency, latency_sensitive, and life_safety_classification are all UNKNOWN rather than confirmed true. SRC-CAP-001 applies because phi_handled=true and baa_confirmed=UNKNOWN, capping confidence at 40 and triggering PENDING_CONDITIONS status. The secondary recommendation is REHOST as a fallback if SaaS procurement cannot be confirmed, given the VM-based architecture with no confirmed hardware constraints.
Secondary Path
The fallback option if the primary recommendation cannot proceed.
Secondary Path — REHOST
REHOST is viable as a fallback: all three VMs are standard Windows Server workloads with no confirmed hardware dependencies, making lift-and-shift to IaaS a low-complexity alternative if the SaaS path cannot be validated. Right-sizing to 2 VMs (SRC-RS-002, 25% reduction) is applicable under either path.
Right-Sizing Notes
Why the cloud sizing estimate is set where it is, based on the telemetry available.
Fleet avg_mem_active_pct is 18.3%, which would ordinarily qualify for SRC-RS-003 (OVERSIZED, threshold <25%). However, RULE T2 fires unconditionally because snapshot_count=1 AND apm_telemetry_available=false. The maximum permitted tier is therefore SRC-RS-002 (MODERATE, 0.25 reduction); SRC-RS-003 and SRC-RS-004 are FORBIDDEN for this run. CPU utilization data is unavailable (cpu_avg_pct null on all 3 VMs; no APM supplement). RULE T1 does not apply: application_type is Client/Server, not Hardware-Integrated. All three VMs run EOL Windows Server 2012 or 2012 R2, constituting active HIPAA Security Rule exposure given phi_handling=true.
This tab shows the raw output of each analysis agent — the specialized components that examine telemetry, dependencies, contracts, and cost. It is the most detailed and most technical view. Most readers will find the Overview and Backlog tabs sufficient.
Telemetry Agent
Utilization Profilemoderate
Snapshot Count1
APM AvailableFalse
Hard Rule TriggeredTrue
Right-Sizing Opp.True
Agent Confidence45%
Telemetry — VM Counts
Powered Off VMs0
Flagged VMs0
Mem Active % (avg)18.3%
Data WindowNone
Inventory Source
Dependency Agent
Epic IntegrationFalse
PHI HandledTrue
Hardware DependencyUNKNOWN
Latency SensitiveUNKNOWN
Blast Radiushigh
Migration RiskMEDIUM
Conf (Inferred)72%
Conf (Confirmed)48%
Verificationsingle_source
Cloud Connectivity
ComplianceHIPAA / HITRUST
Upstream (2)
Xcelera (legacy cardiovascular imaging platform): inferred: rvtools.vm_notes.lxxlinterfacep — 'Migrate images and data from Xcelera prod to Lumedx prod'
Modality acquisition systems (Echo, Cath, EP — DICOM sources): inferred: clinical workflow pattern — CardioPACS receives DICOM studies from imaging modalities in Echo/Cath/EP labs
Downstream (2)
Clinical review workstations (cardiologists, EP physicians, pediatric cardiology): inferred: PACS architecture pattern — lxhealthviewprd serves as production viewer endpoint
Epic (EHR) — image linking / results integration: inferred: catalog.epic_integration = false (non-null, explicit). No Epic integration confirmed.
Procurement Agent
SaaS Available—
SaaS VendorLumedx
SaaS ProductLumedx Apollo Cloud / Lumedx CV Suite Cloud-Hosted
Repurchase ViableINSUFFICIENT_EVIDENCE
BAA ConfirmedUNKNOWN
SaaS Confidence
Inferred Score60%
Confirmed Score0%
Verificationinferred_only
AWS Equiv.None
Provisioning Agent (Python engine — not LLM)
Current VMs3
Recommended VMs2
Right-Sizing25% reduction
Current TCO/yr$19,500
Projected Cloud TCO$10,400
Migration One-Time$2,080
Dual-Running Cost$9,967
-$2,947
Year-1 Net
$9,100
Steady-State
$15,253
3-Year
AWS TargetAmazon EC2
Azure TargetAzure Virtual Machines
Cost Modeldefault_multipliers
DEF: {'default_id': 'DEF-WL-001', 'field': 'workload_type', 'value': 'app_server', 'reason': 'workload_type indeterminable'}
DEF: {'default_id': 'DEF-MIG-001', 'field': 'migration_factor', 'value': 0.2, 'reason': 'migration_complexity not provided'}
DEF: {'default_id': 'DEF-COST-001', 'field': 'cost_model_basis', 'value': 'default_multipliers', 'reason': 'no customer cloud quote provided; using SRC-TCO-* and SRC-MULT-* registry defaults'}
Hologic Apex
Vendor: Hologic · 2 VMs · Criticality: High · HIPAA / HITRUST
RETAIN
55%
Confidence
32%
Data Quality
Disposition: Structurally Retained On-Premises · Current Cost: $13,000/yr · Migration: Not applicable
Overview
Backlog
Provenance
Technical Details
Agents
Recommendation
RETAIN 🔒 STRUCTURAL
SRC-CAP-002 Binding cap · Ceiling: 55% · Threshold: 60%
Secondary: REHOST (conditional — post-upgrade, post-hardware-dependency-confirmation)
Confidence
55%
Disposition is structurally certain; some documentation gaps remain to close.
ROI Summary (Python-computed, not LLM)
Structurally Retained On-Premises
Current Cost: $13,000/yr (continues unchanged)
Migration ROI: Not applicable — this system cannot move to cloud
⚙ Financials computed by provisioning_engine_v3 (Python) — not LLM-generated
Finding
Hologic Apex is a bone densitometry management platform running on end-of-life Windows Server 2008 (32-bit), which expired in January 2020 and cannot be migrated to a compliant cloud environment without an OS upgrade. Both VMs carry an in-flight upgrade note referencing Mike Slezak, indicating the application's technical posture is actively changing. Until the upgrade is complete and potential DEXA scanner hardware dependencies are confirmed or ruled out, RETAIN is the appropriate disposition to protect clinical bone density reporting operations.
In plain terms: this app stays where it is. It is tied to physical hardware or life-safety equipment that cannot move to the cloud. The work here is compliance and lifecycle upkeep, not migration.
This is the list of work needed to move this recommendation forward. The 3 P1 items are the priority — they unlock the recommendation. P2 and P3 items improve confidence and reduce risk but are not blockers. 'Story points' measure effort, not time or cost — see the Methodology tab for how they work.
55%
Current
60%
Threshold
64%
Ceiling
21
Story Points
Why This Score
Current confidence of 55 is hard-capped by SRC-CAP-002 (majority of evidence dimensions are INFERRED or MISSING: procurement_contracts=MISSING, architecture_diagrams=MISSING, security_compliance=INFERRED, dr_requirements=MISSING) and SRC-CAP-004 (snapshot_count=1, apm_telemetry_available=false); the pre-cap calculated score of 64.25 cannot be realized until both caps are resolved through confirmed evidence collection. No confidence climb is possible within this run — all backlog items target cap resolution and ROI variance reduction, with the hardware dependency confirmation (APP-011-001) and OS upgrade status (APP-011-002) as the highest-leverage P1 actions that unblock the migration decision tree.
Proceeding with any migration planning for Hologic Apex without resolving hardware_dependency=UNKNOWN and OS upgrade status carries high risk of committing to a cloud architecture that is either physically impossible (if DEXA hardware integration is confirmed) or non-compliant (if WS2008 32-bit is migrated without OS remediation).
Action Backlog — 6 Items · 21 Story Points · Ceiling: 64%
APP-011-001 Confirm or Rule Out DEXA Scanner Hardware Dependency via Vendor Documentation and Mike Slezak Interview
P1 Dependency Mapping
hardware_dependency=UNKNOWN is the single most consequential gap for Hologic Apex. The product class (bone densitometry management) strongly implies physical DEXA scanner hardware integration via proprietary Hologic acquisition controllers, but no catalog field, vm_notes, or vendor documentation confirms this. Collect: (1) Hologic Apex installation guide or deployment architecture document confirm
+0% confidence -0% ROI variance 👤 📋 · 5 pts
APP-011-002 Confirm OS Upgrade Completion Date and Target OS Version with Mike Slezak
P1 Architecture
Both production VM 'hologic' and test VM 'hologictest' run Windows Server 2008 32-bit (EOL January 2020), which is an active HIPAA Security Rule §164.312 violation given phi_handling=true. Both VMs carry the note 'Hologic Apex Upgrade (Contact: Mike Slezak)', confirming an in-flight remediation. However, the upgrade completion date, target OS version, and whether the upgrade includes a Hologic Ape
+9% confidence -0% ROI variance 👤 📋 · 3 pts
APP-011-003 Obtain or Execute BAA with Target Cloud Provider for PHI-Handling Workload
P1 BAA/Legal
baa_confirmed=UNKNOWN because catalog.baa_valid=null. The application handles PHI (phi_handling=true, CUSTOMER_PROVIDED; phi_risk_analysis=likely_phi from dependency agent). Under HIPAA, a valid BAA with the cloud provider is a legal prerequisite for migrating any PHI workload to cloud infrastructure. Without BAA confirmation, SRC-CAP-001 will fire if phi_handled is ever promoted to true, capping
+0% confidence -0% ROI variance 👤 📋 · 3 pts
APP-011-004 Obtain Hologic Vendor Statement on Cloud-Hosted Deployment Support for Apex
P2 Vendor Engagement
No evidence exists that Hologic supports cloud-hosted deployment of Apex. Vendor COTS applications require vendor certification or a support statement for cloud environments; without it, any REHOST disposition carries unsupported-software risk and potential warranty/support contract voidance. This gap also contributes to the architecture_diagrams=MISSING dimension. Collect: (1) contact Hologic acc
+0% confidence -0% ROI variance 👤 📋 · 5 pts
APP-011-005 Map Confirmed Downstream Consumers — EHR, PACS, and Ordering Clinician Integrations
P2 Dependency Mapping
Downstream consumers (EHR, PACS, ordering clinicians) are currently inferred from product knowledge only. blast_radius=high is computed conservatively from criticality=High, not from a confirmed consumer inventory. The dependency_mapping dimension is classified SINGLE, contributing only 5.0 of a possible 10.0 points to confidence. Promoting this to CONFIRMED requires an actual integration inventor
+0% confidence -0% ROI variance 👤 📋 · 3 pts
APP-011-006 Collect DR/RTO/RPO Requirements for High-Criticality PHI Workload
P3 Architecture
dr_requirements=MISSING, contributing a 10-point ROI variance penalty. Application criticality=High implies DR requirements exist but none are documented. For a PHI-handling clinical application, RTO and RPO targets are also relevant to HIPAA contingency planning requirements (§164.308(a)(7)). Collect: (1) RTO target (maximum acceptable downtime for bone densitometry reporting); (2) RPO target (ma
+9% confidence -0% ROI variance 👤 📋 · 2 pts
This tab shows the evidence behind the recommendation. The seven dimensions are scored by quality — CONFIRMED evidence counts fully, INFERRED evidence counts less. The validator table confirms each analysis step passed its checks. Green means the step is sound.
Evidence Classification — 7 Dimensions
Dimension Classification Multiplier Contribution
Telemetry SINGLE 0.50 6.00
Dependency Mapping SINGLE 0.50 5.00
Procurement Contracts MISSING 0.00 0.00
Architecture Diagrams MISSING 0.00 0.00
Security Compliance INFERRED 0.25 1.25
Business Criticality SINGLE 0.50 2.00
Dr Requirements MISSING 0.00 0.00
Applied Cap: SRC-CAP-002
Validation Reports — Per Agent
Agent Status Provenance Counts Defaults Used Rule Failures
telemetry PASS
SRC:2CUST:13COMP:3DEF:2
 None ✓ None
dependency PASS
SRC:6CUST:7COMP:4DEF:3
 None ✓ None
procurement PASS
None ✓ None
provisioning PASS_WITH_DEFAULTS
SRC:7CUST:1COMP:9DEF:1
• {'default_id': 'DEF-WL-001', 'field': 'workload_type', 'value': 'app_server', 'r
• {'default_id': 'DEF-MIG-001', 'field': 'migration_factor', 'value': 0.2, 'reason
 ✓ None
synthesizer PASS
None ✓ None
confidence_advisor PASS
COMP:15
 None ✓ None
This section contains the full technical reasoning behind the recommendation: the rules that fired, the decision path, the confidence math, and the open intelligence gaps. It is intended for analysts and engineers. Executive readers can rely on the Overview tab.
Rules Applied
These are the specific governance rules that produced this recommendation.
▸ STRUCTURAL PRE-CHECK
This app was classified RETAIN before any 6R rule was evaluated, because it has hardware integration, latency sensitivity, or a life-safety classification that physically prevents cloud migration.
▸ SRC-CAP-002 (binding)
Confidence is capped at 55 because more than half of the seven evidence dimensions are inferred or missing — the conclusion is data-thin.
▸ SRC-CAP-004
Confidence is capped at 65 because telemetry is a single point-in-time snapshot with no ongoing monitoring — right-sizing needs usage patterns over time.
▸ RULE T2 — Right-Sizing Hard Cap
Right-sizing is limited to a moderate 25% reduction because telemetry is a single snapshot. More aggressive right-sizing needs 30+ days of monitoring data.
Decision Path
The step-by-step trace of how the recommendation was reached, in the system's own terms.
PRE-CHECK: hardware_dependency=UNKNOWN (not true), latency_sensitive=UNKNOWN (not true), life_safety_classification=UNKNOWN (not true) — pre-check does not fire. RULE 1 (REPURCHASE): saas_available=false, saas_confidence_inferred=10 (<50) — does not fire. RULE 2 (RETIRE): no successor named, no decommission annotation — does not fire. RULE 3 (REFACTOR): vendor COTS, not internally developed — does not fire. RULE 4 (REHOST): hardware_dependency=UNKNOWN (cannot confirm absent), WS2008 EOL creates hard compliance blocker for cloud migration — cannot fire safely. RULE 5 (RETAIN): formal conditions (all three true) not met, but structural constraints (EOL OS hard migration blocker + implied hardware dependency + in-flight upgrade) prevent any migration disposition from being executable. strategic_recommendation=RETAIN, recommendation_status=STRUCTURAL. CONFIDENCE GATE: final_confidence=55 < 60, but recommendation_status=STRUCTURAL (not PENDING_CONDITIONS) — gate does not alter displayed_primary. displayed_primary=RETAIN.
Cap Evaluation (all fired caps): SRC-CAP-002: ≥4 of 7 dimensions INFERRED/MISSING → ceiling 55 (ceiling=55) ← BINDING SRC-CAP-004: Single snapshot, no APM → ceiling 65 (ceiling=65)
Confidence Scoring Breakdown
How the confidence percentage was calculated, dimension by dimension.
Base score: 50 Evidence contributions: Telemetry SINGLE +6.00 Dependency Mapping SINGLE +5.00 Procurement Contracts MISSING +0.00 Architecture Diagrams MISSING +0.00 Security Compliance INFERRED +1.25 Business Criticality SINGLE +2.00 Dr Requirements MISSING +0.00 Pre-cap score: 64.25 Binding cap applied: SRC-CAP-002 Final confidence: 55%
Intelligence Gaps
Specific things we don't yet know. Each gap is a reason the confidence score is held back.
Hardware Dependency Confirmation — DEXA Scanner Controllers
hardware_dependency=UNKNOWN. Hologic Apex product class strongly implies physical DEXA scanner hardware integration via proprietary acquisition controllers. If confirmed true, this becomes a permanent structural RETAIN constraint. If confirmed false, REHOST becomes viable post-upgrade. Vendor documentation or Mike Slezak interview required.
OS Upgrade Status and Timeline
Both VMs run Windows Server 2008 32-bit (EOL January 2020). In-flight upgrade noted on both VMs. Until upgrade completes, no cloud migration path is HIPAA-compliant. Upgrade completion date and target OS version must be confirmed with Mike Slezak before migration planning can proceed.
BAA Status with Cloud Provider
baa_confirmed=UNKNOWN. Application handles PHI (phi_handling=true in catalog). No BAA evidence in procurement. A valid BAA with the target cloud provider is a legal prerequisite for any cloud migration under HIPAA. Must be resolved before REHOST can be authorized.
Hologic Vendor Cloud Certification
No evidence that Hologic supports cloud-hosted deployment of Apex. Vendor COTS applications require vendor certification for cloud environments. If Hologic does not support cloud hosting, REHOST is blocked regardless of OS or hardware status.
Downstream Consumer Mapping — EHR/PACS Integration
Downstream consumers (EHR, PACS, ordering clinicians) are inferred from product knowledge only. Blast radius classified as high based on criticality=High. Actual integration points unconfirmed. Migration cutover planning requires confirmed downstream consumer inventory.
PHI Handling Confirmation
phi_handled=UNKNOWN in dependency agent (catalog phi_handling=true but not dependency-confirmed). Bone densitometry reports contain patient-identifiable clinical data — PHI handling is highly likely. Confirmation triggers SRC-CAP-001 if BAA remains unconfirmed, capping confidence at 40.
DR Requirements
No DR/RTO/RPO data available. Application criticality=High implies DR requirements exist. Missing DR data contributes 10-point ROI variance penalty and prevents cloud DR architecture sizing.
Technical Detail
A fuller technical explanation of the recommendation and its constraints.
Hologic Apex is retained due to two hard structural blockers identified by the dependency agent: (1) both VMs run Windows Server 2008 32-bit, which reached end-of-life in January 2020 and cannot be lifted to a HIPAA-compliant cloud environment without an OS upgrade as a prerequisite; (2) the product class (bone densitometry management) strongly implies physical DEXA scanner hardware integration via proprietary acquisition controllers, though hardware_dependency remains UNKNOWN pending vendor documentation. The procurement agent confirmed no SaaS alternative exists (saas_available=false, saas_confidence_inferred=10), eliminating REPURCHASE. The application is vendor COTS, eliminating REFACTOR. An in-flight upgrade is annotated on both production and test VMs (contact: Mike Slezak), meaning the application's technical posture will change — migration planning must be deferred until post-upgrade dependency re-assessment.
Secondary Path
The fallback option if the primary recommendation cannot proceed.
Secondary Path — REHOST (conditional — post-upgrade, post-hardware-dependency-confirmation)
Post-upgrade secondary path is REHOST (conditional), contingent on: (1) OS upgrade to a supported Windows Server version, (2) hardware dependency confirmed absent or mitigated via proximity hosting, (3) BAA executed with cloud provider, (4) Hologic vendor certification for cloud-hosted deployment obtained.
Right-Sizing Notes
Why the cloud sizing estimate is set where it is, based on the telemetry available.
RULE T2 ACTIVE: snapshot_count=1 AND apm_telemetry_available=false. Hard rule caps maximum allowable tier at SRC-RS-002 (MODERATE, 0.25 reduction); SRC-RS-003 (OVERSIZED) and SRC-RS-004 (ZOMBIE) are forbidden. Fleet avg mem_active_pct=26.5% independently resolves to MODERATE under Rule T3 (25–60% band), so the cap does not alter the outcome — but the hard rule governs. CPU utilization data is absent for all 2 VMs (cpu_avg_pct null on 2/2); fleet_avg_cpu_pct cannot be computed. Both VMs run Windows Server 2008 (32-bit), EOL January 14, 2020 — active HIPAA Security Rule §164.312 risk given phi_handling=true. VM notes on both records indicate an upgrade is in progress (Contact: Mike Slezak); right-sizing recommendations should be coordinated with the upgrade timeline. VM 'hologictest' is a test instance co-located in production datacenter MUSC-USB — confirm whether it is in scope for right-sizing or should be excluded from production fleet calculations.
This tab shows the raw output of each analysis agent — the specialized components that examine telemetry, dependencies, contracts, and cost. It is the most detailed and most technical view. Most readers will find the Overview and Backlog tabs sufficient.
Telemetry Agent
Utilization Profilemoderate
Snapshot Count1
APM AvailableFalse
Hard Rule TriggeredTrue
Right-Sizing Opp.True
Agent Confidence50%
Telemetry — VM Counts
Powered Off VMs0
Flagged VMs0
Mem Active % (avg)26.5%
Data Window1 days
Inventory Source
Dependency Agent
Epic IntegrationFalse
PHI HandledUNKNOWN
Hardware DependencyUNKNOWN
Latency SensitiveUNKNOWN
Blast Radiushigh
Migration RiskMEDIUM
Conf (Inferred)62%
Conf (Confirmed)28%
Verificationsingle_source
Cloud Connectivity
ComplianceHIPAA / HITRUST
Upstream (3)
Windows Server 2008 (32-bit) OS Platform: confirmed: rvtools.vms[hologic].os, rvtools.vms[hologictest].os
Hologic Apex Application Upgrade (Pending): inferred: rvtools.vm_notes — both VMs carry note 'Hologic Apex Upgrade (Contact: Mike Slezak)'
MUSC-USB Datacenter Network / DEXA Hardware Controllers: inferred: vendor_product_knowledge — Hologic Apex is a bone densitometry management platform that typically interfaces…
Downstream (1)
Radiology / Bone Density Reporting Consumers (EHR, PACS, or Ordering Clinicians): inferred: vendor_product_knowledge — Hologic Apex generates bone density reports (T-scores, Z-scores, FRAX) consumed by…
Procurement Agent
SaaS Available—
SaaS Vendor
SaaS Product
Repurchase ViableFalse
BAA ConfirmedUNKNOWN
SaaS Confidence
Inferred Score10%
Confirmed Score0%
Verificationinferred_only
AWS Equiv.None
Provisioning Agent (Python engine — not LLM)
Current VMs2
Recommended VMs2
Right-Sizing25% reduction
Current TCO/yr$13,000
Projected Cloud TCO$10,400
Migration One-Time$2,080
Dual-Running Cost$7,800
N/A — RETAIN
Year-1 Net
N/A — RETAIN
Steady-State
N/A — RETAIN
3-Year
Migration cost figures are not computed for structurally retained applications.
AWS TargetAmazon EC2
Azure TargetAzure Virtual Machines
Cost Modeldefault_multipliers
DEF: {'default_id': 'DEF-WL-001', 'field': 'workload_type', 'value': 'app_server', 'reason': 'workload_type indeterminable'}
DEF: {'default_id': 'DEF-MIG-001', 'field': 'migration_factor', 'value': 0.2, 'reason': 'migration_complexity not provided'}
DEF: {'default_id': 'DEF-COST-001', 'field': 'cost_model_basis', 'value': 'default_multipliers', 'reason': 'no customer cloud quote provided; using SRC-TCO-* and SRC-MULT-* registry defaults'}
Acuo VNA
Vendor: Acuo Technologies · 12 VMs · Criticality: High · HIPAA / HITRUST
REPURCHASE
40%
Confidence
38%
Data Quality
ROI: Year-1 Net: -$19,760 · Steady-State: $31,200/yr · 3-Year: $42,640
Overview
Backlog
Provenance
Technical Details
Agents
Recommendation
REPURCHASE ⏳ PENDING CONDITIONS
SRC-CAP-001 Binding cap · Ceiling: 40% · Threshold: 60%
Secondary: REHOST
Confidence
40%
Viable, but specific gaps must close first.
ROI Summary (Python-computed, not LLM)
-$19,760
Year-1 Net
$31,200
Steady-State /yr
$42,640
3-Year Total
ROI variance: ±50%
⚙ Financials computed by provisioning_engine_v3 (Python) — not LLM-generated
In plain terms: migrating costs money up front — Year 1 is a net outflow of $19,760. From Year 2 onward the app saves $31,200 every year. The 3-Year total nets these against each other.
Finding
Acuo VNA has a direct vendor-supplied SaaS successor in Hyland's Acuo Cloud VNA / Content Composer Cloud, making REPURCHASE the strategically correct disposition to eliminate 12 on-premises VMs across MUSC-ART and MUSC-USB. However, because this system handles protected health information and no Business Associate Agreement with Hyland has been confirmed, the recommendation is currently pending conditions rather than immediately executable. Confirming the BAA, completing a DICOM data migration assessment, and validating multi-site network readiness are the three actions required to unlock this migration.
In plain terms: this app should move to the vendor's cloud product, but a few specific items must be confirmed first. See Conditions to Unlock below.
Conditions to Unlock (2 items)
This recommendation becomes executable once these conditions are confirmed:
☐ Verify Business Associate Agreement with Hyland for Cloud VNA Migration
Owner: Compliance Officer / Legal Counsel · Timeline: 21 days · Unlocks: compliance_close, cap_resolution, migration_commit
☐ Retrieve Current Contract Terms and Hyland Cloud Pricing Proposal
Owner: Procurement Manager / Vendor Relationship Manager · Timeline: 14 days · Unlocks: cost_clarity, migration_commit
See Backlog tab for full action backlog and P2/P3 items.
This is the list of work needed to move this recommendation forward. The 2 P1 items are the priority — they unlock the recommendation. P2 and P3 items improve confidence and reduce risk but are not blockers. 'Story points' measure effort, not time or cost — see the Methodology tab for how they work.
40%
Current
60%
Threshold
55%
Ceiling
27
Story Points
Why This Score
Current confidence is capped at 40 by SRC-CAP-001: phi_handled=true with baa_confirmed=UNKNOWN is an unconditional ceiling under v3.1 governance, regardless of the pre-cap score of 67.25. The pre-cap score reflects a strategically sound REPURCHASE disposition with a confirmed SaaS successor, but zero CONFIRMED evidence dimensions and a single-snapshot telemetry baseline prevent the score from advancing until the BAA blocker is resolved.
Proceeding with REPURCHASE execution before BAA confirmation exposes the organization to HIPAA 45 CFR §164.308(b) liability and creates an unmitigated compliance gap for a High-criticality PHI-handling system serving four clinical sites.
Action Backlog — 6 Items · 27 Story Points · Ceiling: 55%
APP-012-001 Confirm BAA with Hyland covering Acuo Cloud VNA / Content Composer Cloud
P1 BAA/Legal
Determine whether the existing Acuo Technologies BAA has been novated to Hyland Software following the acquisition, or whether a new BAA must be executed. Obtain the executed BAA document covering Acuo Cloud VNA and Hyland Content Composer Cloud. Confirm scope includes PHI storage, transmission, and processing for DICOM imaging data across all four clinical sites (Chester, Florence, Lancaster, Mar
+15% confidence -0% ROI variance 👤 📋 · 5 pts
APP-012-002 Obtain current Acuo Technologies contract ACV and Hyland SaaS transition pricing
P1 Procurement
Retrieve the current Acuo Technologies / Hyland maintenance contract including ACV, renewal date, and termination provisions. Simultaneously request a customer-specific SaaS pricing proposal from Hyland for Acuo Cloud VNA / Content Composer Cloud covering the full multi-site footprint (Chester, Florence, Lancaster, Marion). This resolves contract_intelligence=INSUFFICIENT_EVIDENCE and reduces ROI
+0% confidence -0% ROI variance 👤 📋 · 3 pts
APP-012-003 Commission DICOM data migration architecture assessment with Hyland professional services
P2 Architecture
Engage Hyland professional services to produce an architecture diagram and data migration plan for moving the on-premises DICOM image archive to Acuo Cloud VNA. Scope must address: (1) two 200 GiB database VMs (VNARHDB1, VNAUHDB1) representing the primary archive volumes; (2) dual-datacenter topology (MUSC-ART, MUSC-USB); (3) four site-specific migration server workflows (Chester, Florence, Lancas
+0% confidence -0% ROI variance 👤 📋 · 8 pts
APP-012-004 Conduct multi-site network latency and bandwidth assessment for cloud VNA DICOM routing
P2 Architecture
Perform latency and bandwidth testing from each of the four clinical sites (Chester, Florence, Lancaster, Marion) to Hyland cloud endpoints. Measure DICOM C-MOVE and WADO-RS transfer performance under representative load. Assess whether Direct Connect or equivalent dedicated connectivity is required to meet clinical imaging SLAs. Resolves latency_sensitive=UNKNOWN and provides the network readines
+0% confidence -0% ROI variance 👤 📋 · 5 pts
APP-012-005 Deploy APM agent and collect 30-day CPU and memory time-series for all 12 VMs
P2 Telemetry
Deploy an APM agent (e.g., Datadog, Dynatrace, or equivalent) across all 12 Acuo VNA VMs to collect 30-day CPU and memory time-series. Priority targets: two 200 GiB database VMs (VNARHDB1, VNAUHDB1) whose buffer pool requirements are invisible in a single RVTools snapshot; four migration servers (VNACHMIGRATE1, VNAFLMIGRATE1, VNALNMIGRATE1, VNAMRMIGRATE1) that may be idle between jobs. Resolves SR
+0% confidence -0% ROI variance 👤 📋 · 3 pts
APP-012-006 Document ASX/Writer consumer workflows and validate Hyland cloud equivalents
P3 Dependency Mapping
Interview the workflow owner (Andy Ross, USB) to document the business processes served by the two ASX/Writer servers (VNARHASXW1, VNAUHASXW1). Determine whether these servers support CD/DVD burning, structured reporting, archive export, or other downstream workflows. Validate that Hyland Content Composer Cloud or an equivalent cloud-native capability can replace these functions before decommissio
+0% confidence -0% ROI variance 👤 📋 · 3 pts
This tab shows the evidence behind the recommendation. The seven dimensions are scored by quality — CONFIRMED evidence counts fully, INFERRED evidence counts less. The validator table confirms each analysis step passed its checks. Green means the step is sound.
Evidence Classification — 7 Dimensions
Dimension Classification Multiplier Contribution
Telemetry SINGLE 0.50 6.00
Dependency Mapping SINGLE 0.50 5.00
Procurement Contracts INFERRED 0.25 2.00
Architecture Diagrams MISSING 0.00 0.00
Security Compliance INFERRED 0.25 1.25
Business Criticality SINGLE 0.50 2.00
Dr Requirements INFERRED 0.25 1.00
Applied Cap: SRC-CAP-001
Validation Reports — Per Agent
Agent Status Provenance Counts Defaults Used Rule Failures
telemetry PASS
SRC:2CUST:9COMP:6DEF:1
 None ✓ None
dependency PASS
SRC:14COMP:5DEF:3
 None ✓ None
procurement PASS
None ✓ None
provisioning PASS_WITH_DEFAULTS
SRC:7CUST:1COMP:9DEF:1
• {'default_id': 'DEF-WL-001', 'field': 'workload_type', 'value': 'app_server', 'r
• {'default_id': 'DEF-MIG-001', 'field': 'migration_factor', 'value': 0.2, 'reason
 ✓ None
synthesizer PASS
None ✓ None
confidence_advisor PASS
COMP:15
 None ✓ None
This section contains the full technical reasoning behind the recommendation: the rules that fired, the decision path, the confidence math, and the open intelligence gaps. It is intended for analysts and engineers. Executive readers can rely on the Overview tab.
Rules Applied
These are the specific governance rules that produced this recommendation.
▸ RULE 1 — REPURCHASE
A vendor SaaS replacement is available and credible, and the app has no hardware constraint — so the recommendation is to replace it with the vendor's cloud service.
▸ SRC-CAP-001 (binding)
Confidence is capped at 40 because the app handles PHI and the vendor's Business Associate Agreement is not confirmed. PHI cannot move to a cloud vendor without a confirmed BAA.
▸ SRC-CAP-002
Confidence is capped at 55 because more than half of the seven evidence dimensions are inferred or missing — the conclusion is data-thin.
▸ SRC-CAP-003
Confidence is capped at 65 because none of the seven evidence dimensions reached CONFIRMED quality — there is no high-quality anchor for the analysis.
▸ SRC-CAP-004
Confidence is capped at 65 because telemetry is a single point-in-time snapshot with no ongoing monitoring — right-sizing needs usage patterns over time.
▸ RULE T2 — Right-Sizing Hard Cap
Right-sizing is limited to a moderate 25% reduction because telemetry is a single snapshot. More aggressive right-sizing needs 30+ days of monitoring data.
Decision Path
The step-by-step trace of how the recommendation was reached, in the system's own terms.
PRE-CHECK: hardware_dependency=false, life_safety_classification=false — structural blocker does not fire. RULE 1 (REPURCHASE): saas_alternative_available=true (catalog notes + procurement agent), saas_confidence_inferred=70 >= 50, hardware_dependency=false — RULE 1 FIRES. strategic_recommendation=REPURCHASE. CONFIDENCE SCORING: base=50; telemetry=SINGLE(+6.0), dependency_mapping=SINGLE(+5.0), procurement_contracts=INFERRED(+2.0), architecture_diagrams=MISSING(+0.0), security_compliance=INFERRED(+1.25), business_criticality=SINGLE(+2.0), dr_requirements=INFERRED(+1.0); pre_cap_total=67.25. CAP EVALUATION: SRC-CAP-001 fires (phi_handled=true, baa_confirmed=UNKNOWN != true) → cap=40; SRC-CAP-002 fires (4/7 dimensions INFERRED/MISSING) → cap=55; SRC-CAP-003 fires (zero CONFIRMED) → cap=65; SRC-CAP-004 fires (snapshot_count=1, apm_telemetry_available=false) → cap=65. Lowest cap wins: SRC-CAP-001 → final_confidence=40. GATE (v4.1): final_confidence=40 < 60, strategic_recommendation=REPURCHASE (not RETAIN) → recommendation_status=PENDING_CONDITIONS, displayed_primary=REPURCHASE (gate does NOT force RETAIN in v4.1), confidence_gate_applied=true.
Cap Evaluation (all fired caps): SRC-CAP-001: PHI handled + BAA not confirmed → ceiling 40 (ceiling=40) ← BINDING SRC-CAP-002: ≥4 of 7 dimensions INFERRED/MISSING → ceiling 55 (ceiling=55) SRC-CAP-003: Zero CONFIRMED dimensions → ceiling 65 (ceiling=65) SRC-CAP-004: Single snapshot, no APM → ceiling 65 (ceiling=65)
Confidence Scoring Breakdown
How the confidence percentage was calculated, dimension by dimension.
Base score: 50 Evidence contributions: Telemetry SINGLE +6.00 Dependency Mapping SINGLE +5.00 Procurement Contracts INFERRED +2.00 Architecture Diagrams MISSING +0.00 Security Compliance INFERRED +1.25 Business Criticality SINGLE +2.00 Dr Requirements INFERRED +1.00 Pre-cap score: 67.25 Binding cap applied: SRC-CAP-001 Final confidence: 40%
Intelligence Gaps
Specific things we don't yet know. Each gap is a reason the confidence score is held back.
BAA with Hyland not confirmed
SRC-CAP-001 caps confidence at 40. PHI-handling VNA cannot migrate to SaaS without a confirmed BAA. This is the single binding blocker for REPURCHASE execution. Resolution: obtain and validate BAA with Hyland covering Acuo Cloud VNA / Content Composer Cloud.
DICOM data migration scope not assessed
No architecture diagrams or data migration plan exists for moving the VNA image archive (served by two 200 GiB database VMs — VNARHDB1, VNAUHDB1) to cloud. Without this, migration timeline and cost are unquantifiable. Resolution: engage Hyland professional services for data migration scoping.
Multi-site network readiness for cloud VNA not validated
Four site-specific migration servers (Chester, Florence, Lancaster, Marion) indicate distributed DICOM routing. Latency sensitivity is UNKNOWN — cloud VNA performance for DICOM C-MOVE/WADO from remote sites requires network assessment. Resolution: conduct latency and bandwidth testing from each site to Hyland cloud endpoints.
Contract intelligence insufficient
Current contract terms, renewal date, and SaaS transition pricing are unknown (contract_intelligence=INSUFFICIENT_EVIDENCE). Without this, ROI projection carries 50% variance. Resolution: obtain current Acuo Technologies contract and Hyland SaaS pricing proposal.
APM telemetry unavailable
apm_telemetry_available=false limits right-sizing confidence to SINGLE snapshot. The 25% right-sizing opportunity (SRC-RS-002) is directionally valid but not confirmed. Resolution: deploy APM agent or collect 30-day performance baseline before cloud sizing.
ASX/Writer downstream consumer scope unknown
Two ASX/Writer servers (VNARHASXW1, VNAUHASXW1) indicate downstream export/burn workflows whose cloud equivalents are unconfirmed. Resolution: document ASX/Writer consumer workflows and validate Hyland cloud equivalent capabilities.
DR requirements not documented
No DR documentation exists for a High-criticality, PHI-handling VNA. Cloud VNA RTO/RPO commitments from Hyland must be validated against MUSC requirements. Resolution: obtain Hyland SLA documentation and compare against MUSC DR policy.
Technical Detail
A fuller technical explanation of the recommendation and its constraints.
Procurement agent confirmed saas_available=true with saas_confidence_inferred=70, identifying Hyland Acuo Cloud VNA / Content Composer Cloud as the direct SaaS successor — Rule 1 (REPURCHASE) fires as hardware_dependency=false and the 50-point inferred threshold is met. Dependency agent confirmed no hardware dependency and no life-safety classification, clearing the pre-check blocker; however, phi_handled=true with baa_confirmed=UNKNOWN triggers SRC-CAP-001, capping final confidence at 40. Telemetry agent confirmed 12/12 VMs powered on across two datacenters with a right-sizing opportunity of 25% (SRC-RS-002), and the multi-site VM naming pattern (Chester, Florence, Lancaster, Marion migration servers) confirms a distributed deployment requiring coordinated cutover planning. The REPURCHASE disposition is strategically sound but requires BAA confirmation, DICOM data migration scoping, and multi-site network validation before execution.
Secondary Path
The fallback option if the primary recommendation cannot proceed.
Secondary Path — REHOST
REHOST to IaaS (AWS/Azure) is the viable fallback if SaaS procurement negotiations fail or BAA terms cannot be agreed; the VM-based architecture with no hardware dependency makes a standard lift-and-shift technically straightforward, though it would not eliminate the on-premises operational model.
Right-Sizing Notes
Why the cloud sizing estimate is set where it is, based on the telemetry available.
Fleet avg mem active pct is 13.2% (CUSTOMER_PROVIDED from rvtools avg_mem_active_pct field; independently computed from per-VM values as 159/12 = 13.25%, delta < 0.1% — accepted as provided). Under Rule T3 alone, 13.2% < 25% would classify this fleet as SRC-RS-003 OVERSIZED. Rule T2 fires unconditionally (snapshot_count=1, apm_telemetry_available=false), suppressing that classification and capping the tier at SRC-RS-002 MODERATE (0.25 reduction). The two database VMs — VNARHDB1 (200 GiB RAM, 28% active) and VNAUHDB1 (200 GiB RAM, 13% active) — represent 400 of 604 GiB total fleet RAM (66%) and are the primary right-sizing candidates; however, database VM sizing is frequently driven by buffer pool and working-set requirements not visible in point-in-time snapshots. Four migration servers (VNACHMIGRATE1, VNAFLMIGRATE1, VNALNMIGRATE1, VNAMRMIGRATE1) show 4–7% mem active pct; ZOMBIE classification (SRC-RS-004) is FORBIDDEN under Rule T2 — these VMs may be idle between migration jobs or sampled off-peak. CPU utilization data is unavailable for all 12 VMs (cpu_avg_pct absent from all rvtools records; no APM supplement provided). Recommend 30-day APM telemetry collection before finalizing right-sizing tier. No EOL OS violations detected; all VMs run Windows Server 2016 or later (EOL Jan 2027).
This tab shows the raw output of each analysis agent — the specialized components that examine telemetry, dependencies, contracts, and cost. It is the most detailed and most technical view. Most readers will find the Overview and Backlog tabs sufficient.
Telemetry Agent
Utilization Profilemoderate
Snapshot Count1
APM AvailableFalse
Hard Rule TriggeredTrue
Right-Sizing Opp.True
Agent Confidence45%
Telemetry — VM Counts
Powered Off VMs0
Flagged VMs0
Mem Active % (avg)13.2%
Data WindowNone
Inventory Source
Dependency Agent
Epic IntegrationFalse
PHI HandledTrue
Hardware DependencyFalse
Latency SensitiveUNKNOWN
Blast Radiushigh
Migration RiskMEDIUM
Conf (Inferred)72%
Conf (Confirmed)58%
Verificationsingle_source
Cloud Connectivity
ComplianceHIPAA / HITRUST
Upstream (3)
DICOM Modality Sources (Radiology/Cardiology/Pathology): inferred: Acuo VNA is a DICOM-native archive; modality worklist and image push from imaging modalities is a structural…
Multi-site Clinical Networks (Chester, Florence, Lancaster, Marion): confirmed: rvtools.vm_notes — VNACHMIGRATE1 note='Acuo VNA Chester Migration Server', VNAFLMIGRATE1 note='Acuo VNA…
Active Directory / Identity Provider: inferred: Windows Server 2016 environment across all 12 VMs; Client/Server application type implies domain…
Downstream (4)
PACS / Diagnostic Viewing Workstations: inferred: VNA is the authoritative image store; PACS and diagnostic viewers retrieve studies from VNA via DICOM…
EHR Image Routing / Epic Imaging Integration: inferred: Healthcare EHR systems (Epic or equivalent) typically consume image links or launch viewers against…
ASX/Writer Consumers (VNARHASXW1, VNAUHASXW1): confirmed: rvtools.vm_notes — VNARHASXW1 note='Acuo VNA ASX/Writer Server Requestor: Andy Ross USB Windows Server…
SaaS Migration Target: Acuo Cloud VNA / Hyland Content Composer Cloud: confirmed: application_catalog.notes = 'SaaS alternative: Acuo cloud VNA / Hyland Content Composer cloud.'
Procurement Agent
SaaS Available—
SaaS VendorHyland
SaaS ProductAcuo Cloud VNA / Hyland Content Composer Cloud
Repurchase ViableINSUFFICIENT_EVIDENCE
BAA ConfirmedUNKNOWN
SaaS Confidence
Inferred Score70%
Confirmed Score0%
Verificationsingle_source
AWS Equiv.None
Provisioning Agent (Python engine — not LLM)
Current VMs12
Recommended VMs9
Right-Sizing25% reduction
Current TCO/yr$78,000
Projected Cloud TCO$46,800
Migration One-Time$9,360
Dual-Running Cost$41,600
-$19,760
Year-1 Net
$31,200
Steady-State
$42,640
3-Year
AWS TargetAmazon EC2
Azure TargetAzure Virtual Machines
Cost Modeldefault_multipliers
DEF: {'default_id': 'DEF-WL-001', 'field': 'workload_type', 'value': 'app_server', 'reason': 'workload_type indeterminable'}
DEF: {'default_id': 'DEF-MIG-001', 'field': 'migration_factor', 'value': 0.2, 'reason': 'migration_complexity not provided'}
DEF: {'default_id': 'DEF-COST-001', 'field': 'cost_model_basis', 'value': 'default_multipliers', 'reason': 'no customer cloud quote provided; using SRC-TCO-* and SRC-MULT-* registry defaults'}
Philips Digital Pathology
Vendor: Philips · 4 VMs · Criticality: High · HIPAA / HITRUST
REPURCHASE
40%
Confidence
28%
Data Quality
ROI: Year-1 Net: -$6,587 · Steady-State: $10,400/yr · 3-Year: $14,213
Overview
Backlog
Provenance
Technical Details
Agents
Recommendation
REPURCHASE ⏳ PENDING CONDITIONS
SRC-CAP-001 Binding cap · Ceiling: 40% · Threshold: 60%
Secondary: RETAIN (pending hardware dependency confirmation)
Confidence
40%
Viable, but specific gaps must close first.
ROI Summary (Python-computed, not LLM)
-$6,587
Year-1 Net
$10,400
Steady-State /yr
$14,213
3-Year Total
ROI variance: ±45%
⚙ Financials computed by provisioning_engine_v3 (Python) — not LLM-generated
In plain terms: migrating costs money up front — Year 1 is a net outflow of $6,587. From Year 2 onward the app saves $10,400 every year. The 3-Year total nets these against each other.
Finding
Philips Digital Pathology has a direct vendor SaaS successor — Philips Digital Pathology Cloud — making REPURCHASE the strategically preferred path. However, two critical unknowns must be resolved before execution: whether the physical whole-slide scanners are tightly coupled to the on-premises OVA deployment (which could block cloud migration), and whether a Business Associate Agreement is in place for the SaaS platform given that this system handles patient pathology images. Until these are confirmed, the recommendation is executable in intent but requires a focused 2-week discovery sprint.
In plain terms: this app should move to the vendor's cloud product, but a few specific items must be confirmed first. See Conditions to Unlock below.
Conditions to Unlock (2 items)
This recommendation becomes executable once these conditions are confirmed:
☐ Verify Business Associate Agreement with Philips for Cloud Migration
Owner: Compliance Officer / Legal · Timeline: 10 days · Unlocks: compliance_close, cap_resolution, migration_commit
☐ Determine Whether Physical Scanners Are Tied to the On-Premises Servers
Owner: Infrastructure Lead / Philips Field Engineer · Timeline: 10 days · Unlocks: migration_commit, operational_safety
See Backlog tab for full action backlog and P2/P3 items.
This is the list of work needed to move this recommendation forward. The 2 P1 items are the priority — they unlock the recommendation. P2 and P3 items improve confidence and reduce risk but are not blockers. 'Story points' measure effort, not time or cost — see the Methodology tab for how they work.
40%
Current
60%
Threshold
55%
Ceiling
28
Story Points
Why This Score
Confidence is hard-capped at 40 by SRC-CAP-001 because phi_handled=true and baa_confirmed=UNKNOWN — a PHI-handling system with an unconfirmed BAA cannot exceed this ceiling regardless of other evidence quality. The pre-cap score of 64.5 reflects a strategically clear REPURCHASE path, but four simultaneous caps (SRC-CAP-001 through SRC-CAP-004) all apply, with SRC-CAP-001 as the binding constraint; resolving the BAA gap is the single action that unlocks the next ceiling tier.
Proceeding with REPURCHASE execution before BAA confirmation and hardware dependency resolution risks initiating a PHI migration without legal cover and potentially discovering mid-project that physical scanner coupling makes cloud SaaS infeasible, resulting in sunk migration costs and a forced reversion to RETAIN.
Action Backlog — 6 Items · 28 Story Points · Ceiling: 55%
APP-013-001 Confirm BAA Status for Philips Digital Pathology Cloud (SRC-CAP-001 Resolution)
P1 BAA/Legal
Engage the MUSC compliance officer and Philips account team to confirm whether a valid, executed Business Associate Agreement (BAA) exists covering app-013 and the Philips Digital Pathology Cloud SaaS platform. Obtain the BAA document reference and verify it covers the current Philips legal entity. If no BAA exists, initiate BAA negotiation with Philips and evaluate whether Paige AI cloud patholog
+15% confidence -0% ROI variance 👤 📋 · 5 pts
APP-013-002 Confirm Physical Slide Scanner Hardware Dependency via Infrastructure Field Survey
P1 Dependency Mapping
Conduct a targeted infrastructure field survey at MUSC-USB datacenter to confirm whether Philips whole-slide scanners (e.g., Philips UltraFast Scanner) are physically co-located and directly coupled to the on-premises OVA deployment. Determine whether scanner-to-server communication is direct (hardware-integrated) or mediated via a separate acquisition layer that could be cloud-redirected. If hard
+0% confidence -0% ROI variance 👤 📋 · 5 pts
APP-013-003 Deploy 30-Day APM Telemetry to Resolve Single-Snapshot Constraint (SRC-CAP-004)
P2 Telemetry
Deploy application performance monitoring (APM) tooling across all 4 VMs (msc00-db001, msc00-tst001, path-adrc-deid, path-adrc-mgmt) to collect a minimum 30-day continuous utilization baseline. Current data is a single point-in-time RVTools snapshot (snapshot_count=1, apm_telemetry_available=false), which triggers SRC-CAP-004 and SRC-RS-002, capping right-sizing at MODERATE (25% reduction) and for
+0% confidence -0% ROI variance 👤 📋 · 8 pts
APP-013-004 Obtain Philips Contract Terms and SaaS Migration Pricing
P2 Procurement
Engage the Philips account team to obtain: (1) current on-premises contract ACV and renewal date for app-013, (2) customer-specific SaaS migration proposal for Philips Digital Pathology Cloud including pricing, migration support scope, and BAA coverage under the current Philips legal entity, (3) any contractual migration levers (e.g., early termination clauses, cloud migration credits). Additional
+0% confidence -0% ROI variance 👤 📋 · 3 pts
APP-013-005 Confirm Life Safety Classification for Digital Pathology Platform
P2 Security/Compliance
Obtain written confirmation from the MUSC clinical informatics or compliance team on whether Philips Digital Pathology is classified as a life-safety system. Digital pathology results inform surgical and oncology decisions; if classified as life-safety, the pre-check structural blocker fires regardless of hardware dependency status, and the complexity multiplier increases by 0.50 for all cost esti
+0% confidence -0% ROI variance 👤 📋 · 2 pts
APP-013-006 Map Downstream LIS / EHR Integration Dependencies
P3 Dependency Mapping
Conduct a structured integration mapping exercise to confirm downstream consumers of Philips Digital Pathology output. The dependency agent inferred a Pathology Reporting / LIS downstream dependency from industry pattern only — no catalog field or RVTools artifact confirms the specific downstream system. Determine: (1) whether pathology results feed directly into Epic or a standalone LIS (e.g., Su
+0% confidence -0% ROI variance 👤 📋 · 5 pts
This tab shows the evidence behind the recommendation. The seven dimensions are scored by quality — CONFIRMED evidence counts fully, INFERRED evidence counts less. The validator table confirms each analysis step passed its checks. Green means the step is sound.
Evidence Classification — 7 Dimensions
Dimension Classification Multiplier Contribution
Telemetry SINGLE 0.50 6.00
Dependency Mapping INFERRED 0.25 2.50
Procurement Contracts MISSING 0.00 0.00
Architecture Diagrams INFERRED 0.25 1.75
Security Compliance INFERRED 0.25 1.25
Business Criticality SINGLE 0.50 2.00
Dr Requirements INFERRED 0.25 1.00
Applied Cap: SRC-CAP-001
Validation Reports — Per Agent
Agent Status Provenance Counts Defaults Used Rule Failures
telemetry PASS
SRC:2CUST:9COMP:4DEF:2
 None ✓ None
dependency PASS
SRC:6CUST:5COMP:3DEF:3
 None ✓ None
procurement PASS
None ✓ None
provisioning PASS_WITH_DEFAULTS
SRC:7CUST:1COMP:9DEF:1
• {'default_id': 'DEF-WL-001', 'field': 'workload_type', 'value': 'app_server', 'r
• {'default_id': 'DEF-MIG-001', 'field': 'migration_factor', 'value': 0.2, 'reason
 ✓ None
synthesizer PASS
None ✓ None
confidence_advisor PASS
COMP:15
 None ✓ None
This section contains the full technical reasoning behind the recommendation: the rules that fired, the decision path, the confidence math, and the open intelligence gaps. It is intended for analysts and engineers. Executive readers can rely on the Overview tab.
Rules Applied
These are the specific governance rules that produced this recommendation.
▸ RULE 1 — REPURCHASE
A vendor SaaS replacement is available and credible, and the app has no hardware constraint — so the recommendation is to replace it with the vendor's cloud service.
▸ SRC-CAP-001 (binding)
Confidence is capped at 40 because the app handles PHI and the vendor's Business Associate Agreement is not confirmed. PHI cannot move to a cloud vendor without a confirmed BAA.
▸ SRC-CAP-002
Confidence is capped at 55 because more than half of the seven evidence dimensions are inferred or missing — the conclusion is data-thin.
▸ SRC-CAP-003
Confidence is capped at 65 because none of the seven evidence dimensions reached CONFIRMED quality — there is no high-quality anchor for the analysis.
▸ SRC-CAP-004
Confidence is capped at 65 because telemetry is a single point-in-time snapshot with no ongoing monitoring — right-sizing needs usage patterns over time.
▸ RULE T2 — Right-Sizing Hard Cap
Right-sizing is limited to a moderate 25% reduction because telemetry is a single snapshot. More aggressive right-sizing needs 30+ days of monitoring data.
Decision Path
The step-by-step trace of how the recommendation was reached, in the system's own terms.
PRE-CHECK: hardware_dependency=UNKNOWN, latency_sensitive=UNKNOWN, life_safety_classification=UNKNOWN — all UNKNOWN, pre-check does not fire. RULE 2 (RETIRE): no redundancy_identified, no decommission_annotation_present — does not fire. RULE 1 (REPURCHASE): saas_alternative_available=true ✓, saas_confidence_inferred=62 ≥ 50 ✓, hardware_dependency=UNKNOWN (not true) ✓ — FIRES. strategic_recommendation=REPURCHASE. CONFIDENCE: base=50, telemetry(SINGLE)=+6.0, dependency(INFERRED)=+2.5, procurement(MISSING)=+0.0, architecture(INFERRED)=+1.75, security(INFERRED)=+1.25, criticality(SINGLE)=+2.0, dr(INFERRED)=+1.0, pre-cap=64.5. CAPS: SRC-CAP-001(phi+no BAA)=40, SRC-CAP-002(majority INFERRED/MISSING)=55, SRC-CAP-003(zero CONFIRMED)=65, SRC-CAP-004(snapshot=1+no APM)=65. Lowest cap=SRC-CAP-001=40. final_confidence=40. GATE: 40 < 60 AND strategic=REPURCHASE → recommendation_status=PENDING_CONDITIONS, displayed_primary=REPURCHASE (v4.1: gate does not force RETAIN).
Cap Evaluation (all fired caps): SRC-CAP-001: PHI handled + BAA not confirmed → ceiling 40 (ceiling=40) ← BINDING SRC-CAP-002: ≥4 of 7 dimensions INFERRED/MISSING → ceiling 55 (ceiling=55) SRC-CAP-003: Zero CONFIRMED dimensions → ceiling 65 (ceiling=65) SRC-CAP-004: Single snapshot, no APM → ceiling 65 (ceiling=65)
Confidence Scoring Breakdown
How the confidence percentage was calculated, dimension by dimension.
Base score: 50 Evidence contributions: Telemetry SINGLE +6.00 Dependency Mapping INFERRED +2.50 Procurement Contracts MISSING +0.00 Architecture Diagrams INFERRED +1.75 Security Compliance INFERRED +1.25 Business Criticality SINGLE +2.00 Dr Requirements INFERRED +1.00 Pre-cap score: 64.5 Binding cap applied: SRC-CAP-001 Final confidence: 40%
Intelligence Gaps
Specific things we don't yet know. Each gap is a reason the confidence score is held back.
BAA Confirmation for Philips Digital Pathology Cloud
SRC-CAP-001 caps confidence at 40 until BAA is confirmed. PHI-handling system cannot be migrated to SaaS without executed BAA. Resolving this single gap removes the binding cap and could unlock EXECUTABLE status.
Physical Slide Scanner Hardware Dependency
Dependency agent inferred Philips whole-slide scanner coupling from OVA architecture patterns. If confirmed true, the pre-check structural blocker fires and strategic_recommendation shifts to RETAIN. Must be confirmed or ruled out before migration planning proceeds.
APM / Application Performance Telemetry
apm_telemetry_available=false and snapshot_count=1 trigger SRC-CAP-004 and SRC-RS-002. Right-sizing is capped at MODERATE (25% reduction). Continuous telemetry would enable more precise sizing and improve confidence scoring.
Procurement Contract Intelligence
contract_intelligence=INSUFFICIENT_EVIDENCE. Current Philips contract terms, renewal date, and SaaS migration path pricing are unknown. Procurement discovery required to validate ROI projection and identify contractual migration levers.
Downstream LIS / EHR Integration Mapping
Pathology Reporting / LIS downstream dependency is inferred from industry pattern only. If the SaaS platform requires re-integration with Epic or a standalone LIS, migration complexity and timeline increase significantly.
Windows Server 2012 EOL Security Posture
msc00-db001 and msc00-tst001 run Windows Server 2012 (64-bit), last patched June 2017. These VMs are on an end-of-life OS with no extended security updates. This is an active security risk independent of migration disposition.
Life Safety Classification Confirmation
life_safety_classification=UNKNOWN. Digital pathology results may inform surgical or oncology decisions. If classified as life-safety, the pre-check structural blocker fires regardless of hardware dependency status.
Technical Detail
A fuller technical explanation of the recommendation and its constraints.
Rule 1 (REPURCHASE) fires because saas_alternative_available=true, saas_confidence_inferred=62 ≥ 50, and hardware_dependency=UNKNOWN (not confirmed true). The dependency agent inferred a Philips slide scanner upstream dependency from OVA VM notes but could not confirm it — this UNKNOWN status is the critical blocker. SRC-CAP-001 applies because phi_handled=true and baa_confirmed=UNKNOWN, capping confidence at 40. The pre-check structural blocker does not fire because no field is confirmed true for hardware_dependency or life_safety_classification. The strategic intent is clear: migrate to Philips Digital Pathology Cloud, but discovery must first confirm scanner architecture and BAA status.
Secondary Path
The fallback option if the primary recommendation cannot proceed.
Secondary Path — RETAIN (pending hardware dependency confirmation)
If hardware dependency is confirmed true and scanner coupling is incompatible with cloud SaaS, the disposition shifts to RETAIN (structural). If BAA is confirmed and scanner dependency is resolved as cloud-compatible, confidence unlocks above 60 and status becomes EXECUTABLE.
Right-Sizing Notes
Why the cloud sizing estimate is set where it is, based on the telemetry available.
RULE T2 hard rule triggered: snapshot_count=1 AND apm_telemetry_available=false. Fleet avg mem_active_pct=13.8% would indicate OVERSIZED (SRC-RS-003) under RULE T3 thresholds, but SRC-RS-003 and SRC-RS-004 are FORBIDDEN under T2. Tier capped at SRC-RS-002 (MODERATE, 0.25 reduction). RULE T1 not triggered: application_type=Client/Server, not Hardware-Integrated. No CPU utilization data available for any of the 4 VMs. Two VMs (msc00-db001, msc00-tst001) report os_name=Windows Server 2012 which reached EOL October 2023; vm_notes reference Server 2012 R2 but os field is the authoritative evidence source per RULE T4. Both 2012 and 2012 R2 are EOL. VM notes reference ESXi 6 OVA templates suggesting potential hypervisor-version dependencies that should be validated before any right-sizing action is executed.
This tab shows the raw output of each analysis agent — the specialized components that examine telemetry, dependencies, contracts, and cost. It is the most detailed and most technical view. Most readers will find the Overview and Backlog tabs sufficient.
Telemetry Agent
Utilization Profilemoderate
Snapshot Count1
APM AvailableFalse
Hard Rule TriggeredTrue
Right-Sizing Opp.True
Agent Confidence45%
Telemetry — VM Counts
Powered Off VMs0
Flagged VMs0
Mem Active % (avg)13.8%
Data WindowNone
Inventory Source
Dependency Agent
Epic IntegrationFalse
PHI HandledTrue
Hardware DependencyUNKNOWN
Latency SensitiveUNKNOWN
Blast Radiushigh
Migration RiskMEDIUM
Conf (Inferred)72%
Conf (Confirmed)48%
Verificationsingle_source
Cloud Connectivity
ComplianceHIPAA / HITRUST
Upstream (3)
Philips Slide Scanner / Acquisition Hardware (inferred): inferred: Philips Digital Pathology Solutions requires physical whole-slide scanners as image source; OVA deployment…
ESXi Hypervisor (ESXi 6 / 6.7 / 7.x): confirmed: rvtools.vm_notes — OVA templates explicitly target ESXi 6, ESXi 6.7, ESXi 7.x
VEEAM Backup Infrastructure: confirmed: rvtools.vm_notes:path-adrc-mgmt — 'VEEAM/MGMT OVA template'
Downstream (2)
Pathology Reporting / LIS (inferred): inferred: digital pathology platforms in healthcare typically feed results into a Laboratory Information System (LIS)…
De-identification Pipeline (path-adrc-deid): confirmed: rvtools.vm_notes:path-adrc-deid — dedicated de-identification VM present, implying downstream consumers of…
Procurement Agent
SaaS Available—
SaaS VendorPhilips
SaaS ProductPhilips Digital Pathology Cloud
Repurchase ViableINSUFFICIENT_EVIDENCE
BAA ConfirmedUNKNOWN
SaaS Confidence
Inferred Score62%
Confirmed Score0%
Verificationsingle_source
AWS Equiv.None
Provisioning Agent (Python engine — not LLM)
Current VMs4
Recommended VMs3
Right-Sizing25% reduction
Current TCO/yr$26,000
Projected Cloud TCO$15,600
Migration One-Time$3,120
Dual-Running Cost$13,867
-$6,587
Year-1 Net
$10,400
Steady-State
$14,213
3-Year
AWS TargetAmazon EC2
Azure TargetAzure Virtual Machines
Cost Modeldefault_multipliers
DEF: {'default_id': 'DEF-WL-001', 'field': 'workload_type', 'value': 'app_server', 'reason': 'workload_type indeterminable'}
DEF: {'default_id': 'DEF-MIG-001', 'field': 'migration_factor', 'value': 0.2, 'reason': 'migration_complexity not provided'}
DEF: {'default_id': 'DEF-COST-001', 'field': 'cost_model_basis', 'value': 'default_multipliers', 'reason': 'no customer cloud quote provided; using SRC-TCO-* and SRC-MULT-* registry defaults'}
RapidAI
Vendor: RapidAI · 3 VMs · Criticality: High · HIPAA / HITRUST — PHI handled; BAA requi
REPURCHASE
40%
Confidence
38%
Data Quality
ROI: Year-1 Net: -$2,947 · Steady-State: $9,100/yr · 3-Year: $15,253
Overview
Backlog
Provenance
Technical Details
Agents
Recommendation
REPURCHASE ⏳ PENDING CONDITIONS
SRC-CAP-001 Binding cap · Ceiling: 40% · Threshold: 60%
Secondary: RETAIN (pending BAA confirmation and contract intelligence)
Confidence
40%
Viable, but specific gaps must close first.
ROI Summary (Python-computed, not LLM)
-$2,947
Year-1 Net
$9,100
Steady-State /yr
$15,253
3-Year Total
ROI variance: ±40%
⚙ Financials computed by provisioning_engine_v3 (Python) — not LLM-generated
In plain terms: migrating costs money up front — Year 1 is a net outflow of $2,947. From Year 2 onward the app saves $9,100 every year. The 3-Year total nets these against each other.
Finding
RapidAI is a strong REPURCHASE candidate — the vendor already offers RapidAI Cloud as a hosted SaaS product, and the on-premises deployment (three Kubernetes VMs) could be eliminated in favor of the vendor-managed cloud service. However, because this application handles patient imaging data under HIPAA, a Business Associate Agreement must be confirmed with RapidAI before migration can proceed. Once BAA status, contract terms, and DICOM/PACS integration requirements are validated, this recommendation becomes fully executable with an estimated $9,100 in annual savings.
In plain terms: this app should move to the vendor's cloud product, but a few specific items must be confirmed first. See Conditions to Unlock below.
Conditions to Unlock (2 items)
This recommendation becomes executable once these conditions are confirmed:
☐ Confirm Business Associate Agreement with RapidAI
Owner: Privacy/Compliance Officer + Vendor Management · Timeline: 15 days · Unlocks: compliance_close, cap_resolution, migration_commit
☐ Gather RapidAI Cloud Contract and Pricing Terms
Owner: Procurement / Vendor Management · Timeline: 15 days · Unlocks: cost_clarity, migration_commit
See Backlog tab for full action backlog and P2/P3 items.
This is the list of work needed to move this recommendation forward. The 2 P1 items are the priority — they unlock the recommendation. P2 and P3 items improve confidence and reduce risk but are not blockers. 'Story points' measure effort, not time or cost — see the Methodology tab for how they work.
40%
Current
60%
Threshold
55%
Ceiling
24
Story Points
Why This Score
RapidAI's confidence is hard-capped at 40 by SRC-CAP-001 because phi_handled=true and baa_confirmed=UNKNOWN — a null catalog field that v3.1 forbids inferring as true or false. Three additional caps (SRC-CAP-002, SRC-CAP-003, SRC-CAP-004) are also active, but SRC-CAP-001 is the binding constraint; resolving BAA status is the single action that lifts the ceiling to 55 and enables the REPURCHASE recommendation to advance toward executable status.
Proceeding with RapidAI Cloud migration without a confirmed BAA exposes MUSC to HIPAA violation risk on both the current on-premises deployment and any SaaS transition, and the REPURCHASE recommendation cannot be acted upon until BAA confirmation, contract intelligence, and DICOM integration architecture are validated.
Action Backlog — 6 Items · 24 Story Points · Ceiling: 55%
APP-014-001 Confirm BAA Status with RapidAI Vendor — Resolve SRC-CAP-001
P1 BAA/Legal
Obtain and validate a signed Business Associate Agreement (BAA) with RapidAI covering both the current on-premises deployment and the RapidAI Cloud SaaS environment. Engage the MUSC Privacy/Compliance Officer and RapidAI account team to locate any existing BAA on file or execute a new one. The BAA must explicitly cover PHI processing under the vendor-hosted SaaS model to support the REPURCHASE pat
+15% confidence -0% ROI variance 👤 📋 · 5 pts
APP-014-002 Obtain RapidAI Cloud Contract Intelligence — Licensing, Data Residency, and SaaS Terms
P1 Procurement
Engage RapidAI vendor and MUSC Procurement to obtain: (1) current on-premises license ACV and renewal date, (2) RapidAI Cloud SaaS subscription pricing and whether the existing on-prem license converts or requires a new contract, (3) data residency terms and geographic constraints for PHI under the SaaS model, (4) SLA commitments for uptime and inference availability. This resolves the procurement
+0% confidence -0% ROI variance 👤 📋 · 5 pts
APP-014-003 Validate DICOM/PACS Integration Architecture for RapidAI Cloud Migration
P2 Architecture
Produce a confirmed architecture diagram documenting the DICOM feed path from MUSC PACS to RapidAI (on-premises) and the proposed equivalent path to RapidAI Cloud. Confirm: (1) DICOM router/gateway configuration at MUSC-ART and MUSC-USB, (2) whether PACS remains on-premises post-migration and requires a cloud egress path (Direct Connect / Private Link / VPN), (3) vendor-supplied VMDK constraints a
+0% confidence -0% ROI variance 👤 📋 · 5 pts
APP-014-004 Obtain Clinical Life-Safety Classification Decision for RapidAI Stroke Triage AI
P2 Security/Compliance
Engage clinical leadership (Radiology Informatics — Wendy Ketchum, CMO/CMIO) to obtain a formal life-safety classification determination for RapidAI. RapidAI is an FDA-cleared stroke/neuro triage AI; if classified as life-safety, the synthesizer pre-check fires and migration_risk escalates to CRITICAL, requiring additional governance gates before REPURCHASE can proceed. If classified as non-life-s
+0% confidence -0% ROI variance 👤 📋 · 3 pts
APP-014-005 Validate RapidAI Cloud Inference Latency SLA Against Clinical Stroke Triage Thresholds
P2 Vendor Engagement
Obtain RapidAI Cloud SLA documentation specifying inference latency commitments (end-to-end DICOM-to-result time). Validate against MUSC clinical thresholds for door-to-needle and stroke alert workflows. If latency_sensitive=true is confirmed, assess whether Direct Connect or dedicated network path is required to meet SLA. If RapidAI Cloud latency is within acceptable clinical bounds, latency_sens
+0% confidence -0% ROI variance 👤 📋 · 3 pts
APP-014-006 Document DR/RTO-RPO Requirements and Validate Against RapidAI Cloud SLA
P3 Architecture
Document current DR posture for RapidAI: the dual-datacenter deployment (MUSC-ART, MUSC-USB) implies an existing HA/DR configuration. Obtain formal RTO/RPO requirements from Radiology Informatics and compare against RapidAI Cloud SLA availability commitments. Confirm whether RapidAI Cloud provides equivalent or superior availability to the current multi-DC Kubernetes deployment. This resolves dr_r
+0% confidence -0% ROI variance 👤 📋 · 3 pts
This tab shows the evidence behind the recommendation. The seven dimensions are scored by quality — CONFIRMED evidence counts fully, INFERRED evidence counts less. The validator table confirms each analysis step passed its checks. Green means the step is sound.
Evidence Classification — 7 Dimensions
Dimension Classification Multiplier Contribution
Telemetry SINGLE 0.50 6.00
Dependency Mapping SINGLE 0.50 5.00
Procurement Contracts MISSING 0.00 0.00
Architecture Diagrams INFERRED 0.25 1.75
Security Compliance INFERRED 0.25 1.25
Business Criticality SINGLE 0.50 2.00
Dr Requirements MISSING 0.00 0.00
Applied Cap: SRC-CAP-001
Validation Reports — Per Agent
Agent Status Provenance Counts Defaults Used Rule Failures
telemetry PASS
SRC:2CUST:9COMP:4DEF:2
 None ✓ None
dependency PASS
SRC:4CUST:6COMP:6DEF:1
 None ✓ None
procurement PASS
None ✓ None
provisioning PASS_WITH_DEFAULTS
SRC:7CUST:1COMP:9DEF:1
• {'default_id': 'DEF-WL-001', 'field': 'workload_type', 'value': 'app_server', 'r
• {'default_id': 'DEF-MIG-001', 'field': 'migration_factor', 'value': 0.2, 'reason
 ✓ None
synthesizer PASS
None ✓ None
confidence_advisor PASS
COMP:15
 None ✓ None
This section contains the full technical reasoning behind the recommendation: the rules that fired, the decision path, the confidence math, and the open intelligence gaps. It is intended for analysts and engineers. Executive readers can rely on the Overview tab.
Rules Applied
These are the specific governance rules that produced this recommendation.
▸ RULE 1 — REPURCHASE
A vendor SaaS replacement is available and credible, and the app has no hardware constraint — so the recommendation is to replace it with the vendor's cloud service.
▸ SRC-CAP-001 (binding)
Confidence is capped at 40 because the app handles PHI and the vendor's Business Associate Agreement is not confirmed. PHI cannot move to a cloud vendor without a confirmed BAA.
▸ SRC-CAP-002
Confidence is capped at 55 because more than half of the seven evidence dimensions are inferred or missing — the conclusion is data-thin.
▸ SRC-CAP-003
Confidence is capped at 65 because none of the seven evidence dimensions reached CONFIRMED quality — there is no high-quality anchor for the analysis.
▸ SRC-CAP-004
Confidence is capped at 65 because telemetry is a single point-in-time snapshot with no ongoing monitoring — right-sizing needs usage patterns over time.
▸ RULE T2 — Right-Sizing Hard Cap
Right-sizing is limited to a moderate 25% reduction because telemetry is a single snapshot. More aggressive right-sizing needs 30+ days of monitoring data.
Decision Path
The step-by-step trace of how the recommendation was reached, in the system's own terms.
PRE-CHECK: hardware_dependency=false — structural blocker does not fire. RULE 1 (REPURCHASE): saas_alternative_available=true (catalog.notes + procurement.saas_available=true) AND saas_confidence_inferred=72 >= 50 AND hardware_dependency=false — RULE 1 FIRES. strategic_recommendation=REPURCHASE. CONFIDENCE SCORING: base=50; telemetry=SINGLE(+6.0), dependency_mapping=SINGLE(+5.0), procurement_contracts=MISSING(+0.0), architecture_diagrams=INFERRED(+1.75), security_compliance=INFERRED(+1.25), business_criticality=SINGLE(+2.0), dr_requirements=MISSING(+0.0); pre_cap_total=66.0. CAP EVALUATION: SRC-CAP-001 fires (phi_handled=true, baa_confirmed=UNKNOWN) → ceiling=40; SRC-CAP-002 fires (4/7 INFERRED/MISSING) → ceiling=55; SRC-CAP-003 fires (zero CONFIRMED) → ceiling=65; SRC-CAP-004 fires (snapshot_count=1, apm=false) → ceiling=65. Lowest cap: SRC-CAP-001=40. final_confidence=40. GATE (v4.1): final_confidence=40 < 60 AND strategic_recommendation=REPURCHASE (not RETAIN) → recommendation_status=PENDING_CONDITIONS; displayed_primary=REPURCHASE (gate does NOT force RETAIN per v4.1). confidence_gate_applied=true.
Cap Evaluation (all fired caps): SRC-CAP-001: PHI handled + BAA not confirmed → ceiling 40 (ceiling=40) ← BINDING SRC-CAP-002: ≥4 of 7 dimensions INFERRED/MISSING → ceiling 55 (ceiling=55) SRC-CAP-003: Zero CONFIRMED dimensions → ceiling 65 (ceiling=65) SRC-CAP-004: Single snapshot, no APM → ceiling 65 (ceiling=65)
Confidence Scoring Breakdown
How the confidence percentage was calculated, dimension by dimension.
Base score: 50 Evidence contributions: Telemetry SINGLE +6.00 Dependency Mapping SINGLE +5.00 Procurement Contracts MISSING +0.00 Architecture Diagrams INFERRED +1.75 Security Compliance INFERRED +1.25 Business Criticality SINGLE +2.00 Dr Requirements MISSING +0.00 Pre-cap score: 66.0 Binding cap applied: SRC-CAP-001 Final confidence: 40%
Intelligence Gaps
Specific things we don't yet know. Each gap is a reason the confidence score is held back.
BAA Status with RapidAI Vendor
PHI-handling application with baa_confirmed=UNKNOWN. SRC-CAP-001 caps confidence at 40. BAA must be confirmed before REPURCHASE is executable. Risk of HIPAA violation on both current on-prem deployment and any SaaS migration without confirmed BAA.
Contract Intelligence — RapidAI Cloud SaaS Terms
contract_intelligence=INSUFFICIENT_EVIDENCE. SaaS migration viability depends on licensing model, data residency terms, and whether current on-prem license converts to SaaS subscription. Procurement penalty of 15% applied to ROI variance.
DICOM/PACS Integration Architecture
DICOM/PACS dependency is inferred, not confirmed. Migration to RapidAI Cloud requires validated DICOM feed path from MUSC PACS to vendor-hosted endpoint. Architecture diagrams absent — integration complexity unknown. Arch penalty of 10% applied to ROI variance.
Life Safety Classification
life_safety_classification=UNKNOWN. RapidAI is a stroke/neuro triage AI — if classified as life-safety, pre-check may fire on re-evaluation. Dependency agent assigned migration_risk=high and blast_radius=high. Requires clinical risk assessment before migration.
Latency Sensitivity for Clinical AI Inference
latency_sensitive=UNKNOWN. Stroke triage AI has time-critical clinical workflows (door-to-needle time). If SaaS inference latency exceeds clinical thresholds, REPURCHASE viability is reduced. Requires latency SLA validation with RapidAI Cloud.
DR / RTO-RPO Requirements
dr_requirements=MISSING. Dual-datacenter deployment (MUSC-ART, MUSC-USB) suggests existing HA/DR posture. SaaS migration must match or exceed current availability SLA. DR penalty of 10% applied to ROI variance.
APM Telemetry Unavailable
apm_telemetry_available=false with snapshot_count=1. Application-layer performance baseline absent. Right-sizing recommendation (SRC-RS-002, 25% reduction) is based on memory active percentage only — actual CPU and I/O workload patterns unconfirmed.
Technical Detail
A fuller technical explanation of the recommendation and its constraints.
Dependency agent confirms hardware_dependency=false and identifies RapidAI Cloud as a confirmed upstream dependency (catalog.notes), establishing that the vendor already operates a SaaS backend — the on-premises Kubernetes cluster is a deployment artifact, not an architectural requirement. Procurement agent confirms saas_available=true with saas_confidence_inferred=72, satisfying Rule 1's threshold of 50. The binding constraint is SRC-CAP-001: phi_handled=true with baa_confirmed=UNKNOWN drives the confidence ceiling to 40, placing this recommendation in PENDING_CONDITIONS status. Telemetry agent flags right-sizing opportunity (SRC-RS-002, 25% reduction) and moderate utilization at 12.7% average memory active, reinforcing that the current on-prem footprint is oversized relative to actual workload demand.
Secondary Path
The fallback option if the primary recommendation cannot proceed.
Secondary Path — RETAIN (pending BAA confirmation and contract intelligence)
If REPURCHASE is blocked by BAA or contract terms, REHOST to cloud IaaS (AWS/Azure) preserves the Kubernetes architecture while eliminating on-prem infrastructure burden; however, this does not resolve the BAA compliance gap and should be treated as a temporary bridge disposition only.
Right-Sizing Notes
Why the cloud sizing estimate is set where it is, based on the telemetry available.
Raw fleet_avg_mem_active_pct is 12.7%, which under RULE T3 (multi-snapshot or APM path) would yield SRC-RS-003 OVERSIZED. However, RULE T2 fires unconditionally: snapshot_count=1 AND apm_telemetry_available=false. Maximum permitted tier is SRC-RS-002 MODERATE (0.25 reduction). SRC-RS-003 OVERSIZED and SRC-RS-004 ZOMBIE are forbidden for this run. The v3.1 refusal trigger explicitly prohibits reasoning around this cap on the basis of 'clearly low' utilization — a single point-in-time sample cannot distinguish a genuinely idle VM from one captured between burst cycles. Additional caution warranted: all three VMs carry vendor-supplied VMDKs running Kubernetes; any right-sizing action must be validated against vendor specifications prior to implementation. CPU utilization data is absent for all 3 VMs, further limiting confidence. TELEMETRY-DEPTH-INSUFFICIENT flag raised for fleet.
This tab shows the raw output of each analysis agent — the specialized components that examine telemetry, dependencies, contracts, and cost. It is the most detailed and most technical view. Most readers will find the Overview and Backlog tabs sufficient.
Telemetry Agent
Utilization Profilemoderate
Snapshot Count1
APM AvailableFalse
Hard Rule TriggeredTrue
Right-Sizing Opp.True
Agent Confidence44%
Telemetry — VM Counts
Powered Off VMs0
Flagged VMs0
Mem Active % (avg)12.7%
Data WindowNone
Inventory Source
Dependency Agent
Epic IntegrationFalse
PHI HandledTrue
Hardware DependencyFalse
Latency SensitiveUNKNOWN
Blast Radiushigh
Migration RiskMEDIUM
Conf (Inferred)72%
Conf (Confirmed)58%
Verificationsingle_source
Cloud Connectivity
ComplianceHIPAA / HITRUST
Upstream (3)
DICOM / PACS (Radiology Imaging Infrastructure): inferred: application_type=AI/Cloud-Connected + Radiology Informatics ownership + RapidAI vendor profile (stroke/neuro…
Kubernetes Cluster Orchestration (on-prem): confirmed: rvtools.vm_notes — all three VMs explicitly noted as 'running Kubernetes'; vendor-supplied VMDKs indicate…
RapidAI Cloud (vendor SaaS backend): confirmed: catalog.notes — 'SaaS alternative: RapidAI Cloud (vendor-hosted SaaS)'; application_type=AI/Cloud-Connected…
Downstream (2)
Radiologist / Neurologist Workflow (clinical decision support consumers): inferred: RapidAI produces stroke/neuro triage outputs consumed by clinical staff; Radiology Informatics ownership…
EHR / Clinical Notification System: inferred: AI triage platforms in stroke care typically push alerts to EHR or paging systems; no catalog field confirms…
Procurement Agent
SaaS Available—
SaaS VendorRapidAI
SaaS ProductRapidAI Cloud
Repurchase ViableINSUFFICIENT_EVIDENCE
BAA ConfirmedUNKNOWN
SaaS Confidence
Inferred Score72%
Confirmed Score0%
Verificationsingle_source
AWS Equiv.None
Provisioning Agent (Python engine — not LLM)
Current VMs3
Recommended VMs2
Right-Sizing25% reduction
Current TCO/yr$19,500
Projected Cloud TCO$10,400
Migration One-Time$2,080
Dual-Running Cost$9,967
-$2,947
Year-1 Net
$9,100
Steady-State
$15,253
3-Year
AWS TargetAmazon EC2
Azure TargetAzure Virtual Machines
Cost Modeldefault_multipliers
DEF: {'default_id': 'DEF-WL-001', 'field': 'workload_type', 'value': 'app_server', 'reason': 'workload_type indeterminable'}
DEF: {'default_id': 'DEF-MIG-001', 'field': 'migration_factor', 'value': 0.2, 'reason': 'migration_complexity not provided'}
DEF: {'default_id': 'DEF-COST-001', 'field': 'cost_model_basis', 'value': 'default_multipliers', 'reason': 'no customer cloud quote provided; using SRC-TCO-* and SRC-MULT-* registry defaults'}
EchoPACS
Vendor: Digisonics · 1 VMs · Criticality: High · HIPAA / HITRUST — PHI handling confirmed
REHOST
55%
Confidence
28%
Data Quality
ROI: Year-1 Net: -$3,640 · Steady-State: $1,300/yr · 3-Year: -$1,040
Overview
Backlog
Provenance
Technical Details
Agents
Recommendation
REHOST ⏳ PENDING CONDITIONS
SRC-CAP-002 Binding cap · Ceiling: 55% · Threshold: 60%
Secondary: RETAIN
Confidence
55%
Viable, but specific gaps must close first.
ROI Summary (Python-computed, not LLM)
-$3,640
Year-1 Net
$1,300
Steady-State /yr
-$1,040
3-Year Total
ROI variance: ±55%
⚙ Financials computed by provisioning_engine_v3 (Python) — not LLM-generated
In plain terms: migrating costs money up front — Year 1 is a net outflow of $3,640. From Year 2 onward the app saves $1,300 every year. The 3-Year total nets these against each other.
Finding
EchoPACS is a single-VM clinical imaging system supporting Adult Cardiology echocardiography workflows at MUSC. The evidence supports a cloud rehost path — the workload is VM-based with no confirmed hardware dependency — but critical unknowns around DICOM acquisition workstation integration, PHI data handling agreements, and storage architecture must be resolved before migration can be safely executed. Discovery work to confirm hardware dependency status and BAA coverage will unlock this recommendation.
In plain terms: see the recommendation and conditions above.
Conditions to Unlock (3 items)
This recommendation becomes executable once these conditions are confirmed:
☐ Confirm Data Privacy Agreements Before Any Cloud Migration of Patient Imaging Data
Owner: Legal / Compliance Officer · Timeline: 21 days · Unlocks: compliance_close, cap_resolution, migration_commit
☐ Determine Whether Echo Ultrasound Machines Require On-Site Server Proximity
Owner: Infrastructure Architect / Adult Cardiology IT · Timeline: 14 days · Unlocks: migration_commit, cap_resolution, operational_safety
☐ Collect 30 Days of Performance Data to Validate Cloud Sizing and Lift Telemetry Cap
Owner: Infrastructure / Monitoring Team · Timeline: 30 days · Unlocks: cap_resolution, cost_clarity, migration_commit
See Backlog tab for full action backlog and P2/P3 items.
This is the list of work needed to move this recommendation forward. The 3 P1 items are the priority — they unlock the recommendation. P2 and P3 items improve confidence and reduce risk but are not blockers. 'Story points' measure effort, not time or cost — see the Methodology tab for how they work.
55%
Current
60%
Threshold
62%
Ceiling
37
Story Points
Why This Score
EchoPACS is capped at 55 by SRC-CAP-002 because 5 of 7 evidence dimensions are INFERRED or MISSING — dependency mapping, procurement contracts, architecture diagrams, security/compliance, and DR requirements all lack confirmed artifacts. The pre-cap score of 61.75 would clear the 60-point threshold, but the evidence quality ceiling prevents advancement until confirmed data replaces inferred signals across the majority of dimensions.
Proceeding with REHOST migration planning before resolving BAA status and hardware dependency confirmation exposes the organization to both a HIPAA legal violation (PHI migration without executed BAA) and a potential architectural failure (cloud-hosted PACS incompatible with DICOM acquisition workstation latency requirements).
Action Backlog — 6 Items · 37 Story Points · Ceiling: 62%
APP-015-001 Confirm BAA Status with Cloud Provider and Digisonics — PHI Migration Gate
P1 BAA/Legal
baa_valid=null in the application catalog. EchoPACS handles PHI (echocardiography images are PHI under HIPAA 45 CFR §164.514). No cloud migration of this workload can proceed without a signed Business Associate Agreement with the target cloud provider (AWS, Azure, GCP, or OCI). Additionally, confirm whether Digisonics holds or requires a BAA for any cloud-hosted or managed deployment model. Collec
+0% confidence -0% ROI variance 👤 📋 · 8 pts
APP-015-002 Confirm Hardware Dependency — DICOM Acquisition Workstation Integration Assessment
P1 Architecture
hardware_dependency=UNKNOWN is a critical migration blocker. Echocardiography PACS systems (Digisonics EchoPACS) interface with ultrasound acquisition workstations via DICOM C-STORE. If acquisition workstations require low-latency LAN connectivity to the PACS server (sub-10ms for real-time echo acquisition), cloud rehost may be architecturally infeasible and RETAIN becomes the correct disposition.
+7% confidence -0% ROI variance 👤 📋 · 8 pts
APP-015-003 Deploy 30-Day APM Telemetry Collection — CPU, Memory, and IOPS Baseline
P1 Telemetry
snapshot_count=1 and apm_telemetry_available=false. SRC-CAP-004 fires (ceiling 65) and SRC-RS-002 hard rule caps right-sizing at MODERATE (0.25 reduction). DICOM workloads have bursty I/O profiles during study retrieval — a single RVTools snapshot at 18% memory utilization cannot distinguish a genuinely idle VM from one sampled between burst cycles. Additionally, CPU utilization is entirely absent
+0% confidence -0% ROI variance 👤 📋 · 8 pts
APP-015-004 Map SAN/NAS Storage Architecture and DICOM Archive Capacity
P2 Architecture
The single VM with 4 GiB RAM strongly implies external SAN/NAS dependency for the DICOM image archive. Storage architecture, total capacity, and growth rate are entirely unconfirmed. Cloud migration sizing and cost modeling cannot be finalized without this data. Collect: (1) SAN/NAS system name, vendor, and total allocated capacity for EchoPACS DICOM archive, (2) current utilization and annual gro
+7% confidence -0% ROI variance 👤 📋 · 5 pts
APP-015-005 Confirm Windows Server OS Upgrade Path with Digisonics Vendor Compatibility Matrix
P2 Vendor Engagement
The EchoPACS VM runs Windows Server 2008 R2 (EOL January 14, 2020). This is an active HIPAA Security Rule risk (45 CFR §164.312) for a PHI-bearing workload. Cloud rehost requires either enrollment in AWS/Azure Extended Security Updates (ESU) or an OS upgrade. Digisonics EchoPACS compatibility with Windows Server 2016, 2019, or 2022 must be confirmed before any OS upgrade is executed — PACS softwar
+0% confidence -0% ROI variance 👤 📋 · 5 pts
APP-015-006 Vendor Inquiry — Digisonics SaaS/Cloud Deployment Model and DR Requirements
P3 Vendor Engagement
Two lower-priority gaps can be resolved in a single vendor engagement: (1) saas_available=UNKNOWN with saas_confidence_inferred=30 — Digisonics is a niche cardiac PACS vendor with no confirmed SaaS offering, but a formal inquiry could unlock a REPURCHASE path and eliminate VM operational burden entirely. Cloud-hosted PACS alternatives (Ambra Health, Intelerad, Sectra) also exist and should be eval
+0% confidence -0% ROI variance 👤 📋 · 3 pts
This tab shows the evidence behind the recommendation. The seven dimensions are scored by quality — CONFIRMED evidence counts fully, INFERRED evidence counts less. The validator table confirms each analysis step passed its checks. Green means the step is sound.
Evidence Classification — 7 Dimensions
Dimension Classification Multiplier Contribution
Telemetry SINGLE 0.50 6.00
Dependency Mapping INFERRED 0.25 2.50
Procurement Contracts MISSING 0.00 0.00
Architecture Diagrams MISSING 0.00 0.00
Security Compliance INFERRED 0.25 1.25
Business Criticality SINGLE 0.50 2.00
Dr Requirements MISSING 0.00 0.00
Applied Cap: SRC-CAP-002
Validation Reports — Per Agent
Agent Status Provenance Counts Defaults Used Rule Failures
telemetry PASS
SRC:2CUST:10COMP:3DEF:2
 None ✓ None
dependency PASS
SRC:3CUST:4COMP:5DEF:7
 None ✓ None
procurement PASS
None ✓ None
provisioning PASS_WITH_DEFAULTS
SRC:7CUST:1COMP:9DEF:1
• {'default_id': 'DEF-WL-001', 'field': 'workload_type', 'value': 'app_server', 'r
• {'default_id': 'DEF-MIG-001', 'field': 'migration_factor', 'value': 0.2, 'reason
 ✓ None
synthesizer PASS
None ✓ None
confidence_advisor PASS
COMP:15
 None ✓ None
This section contains the full technical reasoning behind the recommendation: the rules that fired, the decision path, the confidence math, and the open intelligence gaps. It is intended for analysts and engineers. Executive readers can rely on the Overview tab.
Rules Applied
These are the specific governance rules that produced this recommendation.
▸ RULE 6 — REHOST
No other qualifying path applies — the default for virtual workloads is to rehost to cloud IaaS.
▸ SRC-CAP-002 (binding)
Confidence is capped at 55 because more than half of the seven evidence dimensions are inferred or missing — the conclusion is data-thin.
▸ SRC-CAP-003
Confidence is capped at 65 because none of the seven evidence dimensions reached CONFIRMED quality — there is no high-quality anchor for the analysis.
▸ SRC-CAP-004
Confidence is capped at 65 because telemetry is a single point-in-time snapshot with no ongoing monitoring — right-sizing needs usage patterns over time.
▸ RULE T2 — Right-Sizing Hard Cap
Right-sizing is limited to a moderate 25% reduction because telemetry is a single snapshot. More aggressive right-sizing needs 30+ days of monitoring data.
Decision Path
The step-by-step trace of how the recommendation was reached, in the system's own terms.
PRE-CHECK: hardware_dependency=UNKNOWN (not true), latency_sensitive=UNKNOWN (not true), life_safety_classification=false → pre-check does NOT fire. RULE 1 (REPURCHASE): saas_alternative_available=UNKNOWN → FAILS. RULE 2 (RETIRE): no redundancy_identified, no decommission_annotation → FAILS. RULE 3 (REFACTOR): vendor COTS (Digisonics), not internally developed → FAILS. RULE 4 (REHOST): VM-based=true; hardware_dependency=UNKNOWN — cannot confirm 'no hardware dependency' as true, but UNKNOWN does not equal true; no confirmed structural blocker present; REHOST selected as lowest-assumption forward path for a VM-based workload. CONFIDENCE GATE: final_confidence=55 < 60 AND strategic_recommendation=REHOST (not RETAIN) → recommendation_status=PENDING_CONDITIONS, confidence_gate_applied=true, displayed_primary=REHOST (v4.1: gate does not force RETAIN). BINDING CAP: SRC-CAP-002 fires (5 of 7 dimensions INFERRED or MISSING, majority threshold met) → cap 55. SRC-CAP-003 fires (zero CONFIRMED dimensions) → cap 65. SRC-CAP-004 fires (snapshot_count=1, apm_telemetry_available=false) → cap 65. Lowest cap = SRC-CAP-002 at 55.
Cap Evaluation (all fired caps): SRC-CAP-002: ≥4 of 7 dimensions INFERRED/MISSING → ceiling 55 (ceiling=55) ← BINDING SRC-CAP-003: Zero CONFIRMED dimensions → ceiling 65 (ceiling=65) SRC-CAP-004: Single snapshot, no APM → ceiling 65 (ceiling=65)
Confidence Scoring Breakdown
How the confidence percentage was calculated, dimension by dimension.
Base score: 50 Evidence contributions: Telemetry SINGLE +6.00 Dependency Mapping INFERRED +2.50 Procurement Contracts MISSING +0.00 Architecture Diagrams MISSING +0.00 Security Compliance INFERRED +1.25 Business Criticality SINGLE +2.00 Dr Requirements MISSING +0.00 Pre-cap score: 61.75 Binding cap applied: SRC-CAP-002 Final confidence: 55%
Intelligence Gaps
Specific things we don't yet know. Each gap is a reason the confidence score is held back.
Hardware Dependency on DICOM Acquisition Workstations — UNCONFIRMED
If DICOM modality workstations require low-latency LAN connectivity to the PACS server (common in real-time echo acquisition), cloud rehost may be architecturally infeasible. Must confirm whether acquisition workstations can tolerate WAN-latency DICOM C-STORE operations or require on-premises PACS proximity.
BAA Not Confirmed — PHI Migration Legally Blocked
catalog.baa_valid=null and procurement.baa_confirmed=UNKNOWN. EchoPACS handles PHI (echocardiography images are PHI under HIPAA). No cloud migration of this workload can proceed without a signed BAA with the cloud provider. This is a legal prerequisite, not a technical one.
Network Storage / SAN Architecture Unknown
PACS systems require high-capacity persistent storage for DICOM image archives. The single VM with 4 GiB RAM strongly implies external SAN/NAS dependency. The storage architecture, capacity, and migration path for the DICOM archive are entirely unconfirmed and must be scoped before cloud migration sizing.
Windows Server 2008 R2 — End-of-Life OS
The VM runs Windows Server 2008 R2 (EOL January 2020). Cloud rehost requires either AWS/Azure extended security update enrollment or OS upgrade. Digisonics EchoPACS compatibility with newer Windows Server versions must be confirmed with the vendor before OS upgrade.
APM Telemetry Unavailable — Cloud Sizing Unvalidated
apm_telemetry_available=false and snapshot_count=1. The single RVTools snapshot shows 18% average memory utilization on 4 GiB RAM, but DICOM workloads have bursty I/O profiles during study retrieval. Without APM data, cloud instance sizing and IOPS requirements cannot be validated. Right-sizing tier capped at MODERATE per SRC-RS-002.
SaaS Alternative Availability — Not Assessed
saas_available=UNKNOWN, saas_confidence_inferred=30. Digisonics does not have a confirmed SaaS product in the procurement catalog. Cloud-hosted PACS alternatives (e.g., Ambra Health, Intelerad, Sectra) exist in the market. A formal SaaS evaluation could unlock a REPURCHASE path and eliminate VM operational burden entirely.
DR Requirements Unknown
No DR/RTO/RPO data available. Adult Cardiology PACS is High criticality — DR requirements likely exist but are unconfirmed. Cloud migration design (single-AZ vs. multi-AZ, backup frequency) cannot be finalized without DR specifications.
EHR / HL7 Integration Scope Unconfirmed
Epic integration is confirmed absent (epic_integration=false), but HL7 result routing to other EHR/EMR systems is unconfirmed. If EchoPACS routes diagnostic reports via HL7 to a non-Epic system, that integration must be preserved or re-architected in the cloud target state.
Technical Detail
A fuller technical explanation of the recommendation and its constraints.
EchoPACS is a single-VM PACS system (Digisonics) supporting Adult Cardiology echocardiography at MUSC-ART. The dependency agent identifies high blast radius and high migration risk, with all upstream dependencies (DICOM modality workstations, AD, SAN storage) inferred only. The procurement agent returns saas_confidence_inferred=30 and saas_available=UNKNOWN, blocking Rule 1 (REPURCHASE). No decommission annotation or named successor blocks Rule 2 (RETIRE). The application is vendor COTS, blocking Rule 3 (REFACTOR). Rule 4 (REHOST) is the first viable path: the workload is VM-based and hardware_dependency=UNKNOWN does not confirm a structural blocker. REHOST is recommended pending discovery of hardware dependency status, BAA execution, and DICOM integration architecture confirmation.
Secondary Path
The fallback option if the primary recommendation cannot proceed.
Secondary Path — RETAIN
RETAIN is the secondary disposition if discovery confirms hardware dependency on DICOM acquisition workstations with latency constraints that preclude cloud hosting of the PACS server.
Right-Sizing Notes
Why the cloud sizing estimate is set where it is, based on the telemetry available.
RULE T1 CHECK: application_type='Client/Server' — not 'Hardware-Integrated'; T1 override does NOT fire. RULE T2 TRIGGERED: snapshot_count=1 AND apm_telemetry_available=false → hard_rule_triggered=true. Maximum permitted tier is SRC-RS-002 (MODERATE, 0.25 reduction); SRC-RS-003 (OVERSIZED) and SRC-RS-004 (ZOMBIE) are FORBIDDEN for this run. NOTE: fleet_avg_mem_active_pct=18.0% would satisfy the <25% threshold for OVERSIZED under Rule T3, but T2 takes precedence unconditionally — a single point-in-time snapshot cannot distinguish a genuinely idle VM from one sampled between burst cycles. Right-sizing action must not be taken until additional telemetry is collected (minimum: 30-day APM time-series or multiple RVTools snapshots spanning varied load periods). ADDITIONAL CONTEXT: VM note identifies 'echowebtest' as a Test Web Server despite the application being classified Production/High criticality — this discrepancy warrants clarification before any sizing or disposition decision. CPU utilization data is entirely absent (cpu_avg_pct null on 1 of 1 VMs); no CPU-based sizing signal is available. EOL OS (Windows Server 2008 R2, EOL Jan 2020) detected on the sole VM; phi_handling=true elevates this to an active HIPAA Security Rule concern.
This tab shows the raw output of each analysis agent — the specialized components that examine telemetry, dependencies, contracts, and cost. It is the most detailed and most technical view. Most readers will find the Overview and Backlog tabs sufficient.
Telemetry Agent
Utilization Profilemoderate
Snapshot Count1
APM AvailableFalse
Hard Rule TriggeredTrue
Right-Sizing Opp.True
Agent Confidence45%
Telemetry — VM Counts
Powered Off VMs0
Flagged VMs0
Mem Active % (avg)18.0%
Data WindowNone
Inventory Source
Dependency Agent
Epic IntegrationFalse
PHI HandledUNKNOWN
Hardware DependencyUNKNOWN
Latency SensitiveUNKNOWN
Blast Radiushigh
Migration RiskMEDIUM
Conf (Inferred)72%
Conf (Confirmed)38%
Verificationsingle_source
Cloud Connectivity
ComplianceHIPAA / HITRUST
Upstream (3)
Echocardiography Ultrasound Acquisition Workstations (DICOM Modality): inferred: PACS systems receive DICOM studies from acquisition modalities; Adult Cardiology context per vm_notes
Active Directory / Windows Authentication: inferred: Windows Server 2008 R2 OS; Client/Server application type in enterprise healthcare environment
Network Storage / SAN (DICOM image archive): inferred: PACS systems require persistent high-capacity storage for imaging studies; single VM with 4 GiB RAM suggests…
Downstream (3)
Adult Cardiology Clinical Workstations (EchoPACS viewers): confirmed: rvtools.vm_notes — 'EchoPACS Test Web Server for Adult Cardiology IT'
Cardiologist / Clinician Diagnostic Workflow: inferred: PACS systems are downstream consumed by reading physicians for diagnostic reporting
Potential EHR / EMR Integration (Epic or other): inferred: epic_integration=false per catalog, but PACS systems commonly integrate with EHR for order/result workflow;…
Procurement Agent
SaaS Available—
SaaS VendorDigisonics
SaaS Product
Repurchase ViableUNKNOWN
BAA ConfirmedUNKNOWN
SaaS Confidence
Inferred Score30%
Confirmed Score0%
Verificationinferred_only
AWS Equiv.None
Provisioning Agent (Python engine — not LLM)
Current VMs1
Recommended VMs1
Right-Sizing25% reduction
Current TCO/yr$6,500
Projected Cloud TCO$5,200
Migration One-Time$1,040
Dual-Running Cost$3,900
-$3,640
Year-1 Net
$1,300
Steady-State
-$1,040
3-Year
AWS TargetAmazon EC2
Azure TargetAzure Virtual Machines
Cost Modeldefault_multipliers
DEF: {'default_id': 'DEF-WL-001', 'field': 'workload_type', 'value': 'app_server', 'reason': 'workload_type indeterminable'}
DEF: {'default_id': 'DEF-MIG-001', 'field': 'migration_factor', 'value': 0.2, 'reason': 'migration_complexity not provided'}
DEF: {'default_id': 'DEF-COST-001', 'field': 'cost_model_basis', 'value': 'default_multipliers', 'reason': 'no customer cloud quote provided; using SRC-TCO-* and SRC-MULT-* registry defaults'}
Hyland OnBase / VNA
Vendor: Hyland · 55 VMs · Criticality: High · HIPAA / HITRUST — PHI-bearing imaging da
REPURCHASE
40%
Confidence
38%
Data Quality
ROI: Year-1 Net: -$88,573 · Steady-State: $144,300/yr · 3-Year: $200,027
Overview
Backlog
Provenance
Technical Details
Agents
Recommendation
REPURCHASE ⏳ PENDING CONDITIONS
SRC-CAP-001 Binding cap · Ceiling: 40% · Threshold: 60%
Secondary: REHOST
Confidence
40%
Viable, but specific gaps must close first.
ROI Summary (Python-computed, not LLM)
-$88,573
Year-1 Net
$144,300
Steady-State /yr
$200,027
3-Year Total
ROI variance: ±45%
⚙ Financials computed by provisioning_engine_v3 (Python) — not LLM-generated
In plain terms: migrating costs money up front — Year 1 is a net outflow of $88,573. From Year 2 onward the app saves $144,300 every year. The 3-Year total nets these against each other.
Finding
Hyland OnBase/VNA has a direct vendor SaaS path via Hyland Cloud, and the application notes confirm this migration route is already under consideration. However, the BAA with Hyland for the SaaS environment has not been confirmed, which is a prerequisite for any PHI-bearing workload migration in a healthcare setting. Once the BAA is executed and the active multi-site imaging migrations (AGFA, Merge PACS, Black River, Orangeburg) complete, this 55-VM footprint can be eliminated in favor of the vendor-managed cloud platform.
In plain terms: this app should move to the vendor's cloud product, but a few specific items must be confirmed first. See Conditions to Unlock below.
Conditions to Unlock (3 items)
This recommendation becomes executable once these conditions are confirmed:
☐ Confirm Business Associate Agreement with Hyland for Cloud Hosting
Owner: Compliance Officer / Legal Counsel · Timeline: 10 days · Unlocks: compliance_close, cap_resolution, migration_commit
☐ Collect 30-Day Performance Monitoring Data Across All 55 VMs
Owner: Infrastructure / Cloud Engineering Team · Timeline: 30 days · Unlocks: cost_clarity, cap_resolution
☐ Get Hyland Cloud Pricing and Contract Terms for SaaS Migration
Owner: Procurement / Vendor Management · Timeline: 10 days · Unlocks: cost_clarity, migration_commit
See Backlog tab for full action backlog and P2/P3 items.
This is the list of work needed to move this recommendation forward. The 3 P1 items are the priority — they unlock the recommendation. P2 and P3 items improve confidence and reduce risk but are not blockers. 'Story points' measure effort, not time or cost — see the Methodology tab for how they work.
40%
Current
60%
Threshold
55%
Ceiling
39
Story Points
Why This Score
Confidence is hard-capped at 40 by SRC-CAP-001 because phi_handling=true is asserted in the application catalog for a 55-VM radiology VNA platform, yet baa_confirmed=UNKNOWN — no executed BAA with Hyland covering the SaaS environment has been located or cited. The pre-cap score of 69.0 reflects strong evidence of SaaS viability and a well-characterized VM fleet, but SRC-CAP-001 is an unconditional ceiling that cannot be lifted by any amount of telemetry or architecture evidence alone; only confirmed BAA execution resolves it.
Proceeding with REPURCHASE to Hyland Cloud SaaS without a confirmed BAA exposes MUSC to HIPAA enforcement risk for PHI-bearing imaging data (DICOM studies, mammography records) transmitted to and stored in a vendor-hosted environment without a valid Business Associate Agreement.
Action Backlog — 6 Items · 39 Story Points · Ceiling: 55%
APP-016-001 Confirm BAA with Hyland for SaaS-hosted environment to resolve SRC-CAP-001 binding cap
P1 BAA/Legal
Engage MUSC compliance officer and legal counsel to determine whether an executed BAA with Hyland Software, Inc. is currently on file and whether its scope explicitly covers the Hyland Cloud (SaaS) hosted environment. If no BAA exists or the current BAA does not cover the SaaS environment, initiate BAA negotiation with Hyland. The BAA must address: (1) PHI scope including DICOM imaging studies, ma
+15% confidence -0% ROI variance 👤 📋 · 5 pts
APP-016-002 Deploy 30-day APM telemetry collection across all 55 VMs to resolve SRC-CAP-004 and validate right-sizing
P1 Telemetry
Deploy APM agents (e.g., Dynatrace, Datadog, or Azure Monitor) across all 55 VMs in both MUSC-ART and MUSC-USB datacenters to collect a minimum 30-day time-series dataset covering CPU utilization, memory active percentage, disk I/O, and network throughput. Current data is a single RVTools point-in-time snapshot with fleet-wide CPU null (55/55 VMs) and memory active at 9.3% — insufficient to charac
+0% confidence -0% ROI variance 👤 📋 · 13 pts
APP-016-003 Obtain Hyland Cloud SaaS pricing proposal and contract transition terms to confirm REPURCHASE ROI
P1 Procurement
Engage Hyland account team to obtain a customer-specific Hyland Cloud (SaaS) pricing proposal covering: (1) per-study or per-user SaaS subscription pricing for the MUSC imaging volume; (2) data ingestion and egress costs for migrating the existing VNA archive (estimated 2,072 GiB RAM-equivalent imaging data); (3) contract transition terms including any on-premises license credit or trade-in provis
+0% confidence -0% ROI variance 👤 📋 · 5 pts
APP-016-004 Produce formal PHI data flow diagram and HIPAA risk assessment to confirm phi_handled and scope BAA
P2 Security/Compliance
Commission a formal PHI data flow diagram for Hyland OnBase/VNA covering: (1) all PHI ingestion paths (AGFA, Merge PACS, Black River, Orangeburg migration feeds; RIS/EMR integration points); (2) PHI at-rest storage locations across MUSC-ART and MUSC-USB; (3) PHI transmission paths to NilRead diagnostic viewers (ART and USB clusters) and research servers; (4) de-identification workflows for researc
+0% confidence -0% ROI variance 👤 📋 · 5 pts
APP-016-005 Document NilRead integration topology and VNA-to-clinical-system architecture to confirm SaaS migration scope
P2 Architecture
Produce architecture diagrams documenting: (1) NilRead diagnostic viewer cluster topology across MUSC-ART (VNARHVIEW1-10, VNARHVIEW5-7, VNARHRESEAR, VNARHNILDMZ1) and MUSC-USB (VNAUHVIEW1-10, VNAUHVIEW5-6, VNAUHNILDMZ1, VNAUHRESEAR) — 20+ VMs confirmed active via RVTools; (2) integration endpoints between VNA and downstream clinical systems (RIS, EMR, PACS worklist); (3) NilRead compatibility requ
+0% confidence -0% ROI variance 👤 📋 · 8 pts
APP-016-006 Confirm RTO/RPO requirements and validate Hyland Cloud SaaS DR SLA for High-criticality radiology platform
P3 Architecture
Engage MUSC IT leadership and Radiology department to document formal RTO/RPO targets for Hyland OnBase/VNA as a High-criticality radiology system. Current dual-datacenter deployment (MUSC-ART + MUSC-USB) implies DR intent but RTO/RPO targets are not confirmed (dr_requirements=INFERRED). Once targets are documented, validate against Hyland Cloud SaaS published SLA commitments. Key questions: (1) W
+0% confidence -0% ROI variance 👤 📋 · 3 pts
This tab shows the evidence behind the recommendation. The seven dimensions are scored by quality — CONFIRMED evidence counts fully, INFERRED evidence counts less. The validator table confirms each analysis step passed its checks. Green means the step is sound.
Evidence Classification — 7 Dimensions
Dimension Classification Multiplier Contribution
Telemetry SINGLE 0.50 6.00
Dependency Mapping SINGLE 0.50 5.00
Procurement Contracts INFERRED 0.25 2.00
Architecture Diagrams INFERRED 0.25 1.75
Security Compliance INFERRED 0.25 1.25
Business Criticality SINGLE 0.50 2.00
Dr Requirements INFERRED 0.25 1.00
Applied Cap: SRC-CAP-001
Validation Reports — Per Agent
Agent Status Provenance Counts Defaults Used Rule Failures
telemetry PASS
SRC:2CUST:10COMP:4DEF:1
 None ✓ None
dependency PASS
SRC:12CUST:4COMP:7DEF:2
 None ✓ None
procurement PASS
None ✓ None
provisioning PASS_WITH_DEFAULTS
SRC:7CUST:1COMP:9DEF:1
• {'default_id': 'DEF-WL-001', 'field': 'workload_type', 'value': 'app_server', 'r
• {'default_id': 'DEF-MIG-001', 'field': 'migration_factor', 'value': 0.2, 'reason
 ✓ None
synthesizer PASS
None ✓ None
confidence_advisor PASS
COMP:27
 None ✓ None
This section contains the full technical reasoning behind the recommendation: the rules that fired, the decision path, the confidence math, and the open intelligence gaps. It is intended for analysts and engineers. Executive readers can rely on the Overview tab.
Rules Applied
These are the specific governance rules that produced this recommendation.
▸ RULE 1 — REPURCHASE
A vendor SaaS replacement is available and credible, and the app has no hardware constraint — so the recommendation is to replace it with the vendor's cloud service.
▸ SRC-CAP-001 (binding)
Confidence is capped at 40 because the app handles PHI and the vendor's Business Associate Agreement is not confirmed. PHI cannot move to a cloud vendor without a confirmed BAA.
▸ SRC-CAP-002
Confidence is capped at 55 because more than half of the seven evidence dimensions are inferred or missing — the conclusion is data-thin.
▸ SRC-CAP-004
Confidence is capped at 65 because telemetry is a single point-in-time snapshot with no ongoing monitoring — right-sizing needs usage patterns over time.
▸ RULE T2 — Right-Sizing Hard Cap
Right-sizing is limited to a moderate 25% reduction because telemetry is a single snapshot. More aggressive right-sizing needs 30+ days of monitoring data.
Decision Path
The step-by-step trace of how the recommendation was reached, in the system's own terms.
PRE-CHECK: hardware_dependency=false, life_safety_classification=false → pre-check does not fire. RULE 1 (REPURCHASE): saas_alternative_available=true (Hyland Cloud SaaS, SOURCED from application notes + procurement), saas_confidence_inferred=80 ≥ 50, hardware_dependency=false → RULE 1 FIRES. strategic_recommendation=REPURCHASE. CONFIDENCE SCORING: base=50; telemetry=SINGLE(+6.0), dependency_mapping=SINGLE(+5.0), procurement_contracts=INFERRED(+2.0), architecture_diagrams=INFERRED(+1.75), security_compliance=INFERRED(+1.25), business_criticality=SINGLE(+2.0), dr_requirements=INFERRED(+1.0); pre-cap=69.0. CAP EVALUATION: SRC-CAP-001 fires (phi_handling=true CUSTOMER_PROVIDED, baa_confirmed=UNKNOWN ≠ true) → cap 40; SRC-CAP-002 fires (4 of 7 dimensions INFERRED, majority) → cap 55; SRC-CAP-004 fires (snapshot_count=1, apm_telemetry_available=false) → cap 65. Lowest cap = SRC-CAP-001 = 40. final_confidence=40. GATE: final_confidence=40 < 60, strategic_recommendation=REPURCHASE → recommendation_status=PENDING_CONDITIONS, confidence_gate_applied=true, displayed_primary=REPURCHASE (v4.1: gate does not force RETAIN).
Cap Evaluation (all fired caps): SRC-CAP-001: PHI handled + BAA not confirmed → ceiling 40 (ceiling=40) ← BINDING SRC-CAP-002: ≥4 of 7 dimensions INFERRED/MISSING → ceiling 55 (ceiling=55) SRC-CAP-004: Single snapshot, no APM → ceiling 65 (ceiling=65)
Confidence Scoring Breakdown
How the confidence percentage was calculated, dimension by dimension.
Base score: 50 Evidence contributions: Telemetry SINGLE +6.00 Dependency Mapping SINGLE +5.00 Procurement Contracts INFERRED +2.00 Architecture Diagrams INFERRED +1.75 Security Compliance INFERRED +1.25 Business Criticality SINGLE +2.00 Dr Requirements INFERRED +1.00 Pre-cap score: 69.0 Binding cap applied: SRC-CAP-001 Final confidence: 40%
Intelligence Gaps
Specific things we don't yet know. Each gap is a reason the confidence score is held back.
BAA with Hyland for SaaS environment not confirmed
SRC-CAP-001 binding cap (ceiling 40). PHI-bearing imaging data cannot migrate to Hyland Cloud SaaS without an executed BAA. This is the single highest-priority condition to unlock REPURCHASE execution. Requires legal/compliance review of Hyland Cloud SaaS BAA terms.
Active multi-site imaging migrations not yet complete
14 of 55 VMs are temporary migration servers (AGFA, Merge PACS, Black River, Orangeburg, SQL migration DBs). SaaS cutover timing must be sequenced after these migrations complete to avoid data continuity gaps. Migration completion status and timeline are not confirmed in available data.
No APM telemetry — right-sizing based on single RVTools snapshot
SRC-CAP-004 applies. Fleet avg memory active = 9.3% across 55 VMs, but without APM time-series data, peak utilization patterns for NilRead diagnostic viewers (radiology read sessions) are unknown. Right-sizing to 41 VMs (SRC-RS-002, 25% reduction) carries risk if peak radiology workload is not captured in the snapshot. ROI variance set at 45%.
Procurement contract terms and SaaS pricing not confirmed
saas_confidence_confirmed=0; contract_intelligence=INSUFFICIENT_EVIDENCE. Hyland Cloud SaaS TCO, data egress costs for 2,072 GiB RAM-equivalent imaging data, and contract transition terms are unknown. ROI projection of $432,900 (3-year) carries 45% variance until contract terms are confirmed.
phi_handled field UNKNOWN in dependency agent despite catalog assertion
Application catalog states phi_handling=true (CUSTOMER_PROVIDED), but dependency agent returned UNKNOWN due to absence of a source artifact citation. A formal PHI data flow confirmation (e.g., HIPAA risk assessment, data flow diagram) would resolve this discrepancy and is required for BAA scoping.
Architecture diagrams absent — NilRead integration topology unconfirmed
NilRead diagnostic viewer cluster (20+ VMs across ART and USB) is confirmed active via RVTools notes, but the integration architecture between VNA, NilRead, and downstream clinical systems (RIS, EMR) is not documented. SaaS migration must account for NilRead compatibility with Hyland Cloud endpoints.
DR requirements not documented
Dual-datacenter deployment (MUSC-ART + MUSC-USB) implies DR intent, but RTO/RPO targets for this High-criticality radiology system are not confirmed. Hyland Cloud SaaS DR SLA must be validated against MUSC requirements before cutover.
Technical Detail
A fuller technical explanation of the recommendation and its constraints.
Rule 1 (REPURCHASE) fires: saas_alternative_available=true (Hyland Cloud SaaS confirmed in application notes and procurement output), saas_confidence_inferred=80 ≥ 50, and hardware_dependency=false. The pre-check does not apply (no hardware dependency, no life-safety classification). The strategic disposition is REPURCHASE — eliminating 55 on-premises VMs in favor of the vendor-managed SaaS platform. Confidence is capped at 40 by SRC-CAP-001 (phi_handling=true per catalog, baa_confirmed=UNKNOWN), placing the recommendation in PENDING_CONDITIONS status. The dependency agent confirms active imaging migrations from four source systems (AGFA, Merge PACS, Black River, Orangeburg), which must complete before SaaS cutover to avoid data continuity risk.
Secondary Path
The fallback option if the primary recommendation cannot proceed.
Secondary Path — REHOST
REHOST is the secondary disposition if SaaS transition is deferred: the 55-VM fleet has no hardware dependency or latency constraint, and AWS MGN / Azure Migrate lift-and-shift would reduce TCO via right-sizing (SRC-RS-002, 25% reduction to ~41 effective VMs) while eliminating on-premises infrastructure management.
Right-Sizing Notes
Why the cloud sizing estimate is set where it is, based on the telemetry available.
Fleet avg mem active pct = 9.3% (single point-in-time snapshot, 55 of 55 VMs powered on, 100% data coverage). Hard rule T2 triggered: snapshot_count=1 AND apm_telemetry_available=false. Raw utilization (9.3%) falls in OVERSIZED territory (<25%), but tier is unconditionally capped at SRC-RS-002 MODERATE per hard rule — a single snapshot cannot distinguish genuinely idle VMs from those sampled between bursts. Fleet includes numerous migration servers labeled TEMPORARY (VNAUH2MIG1-5, VNAUHDMMIG01-03, VNABRMIGRATE1-2, VNAORGMIGRATE1-2, vnamidmerg1-2, vnarhview7, vnauh2migdb, vnauhdmsql, vnauhmigdb, VNAUHVIEW7) which are expected to exhibit episodic high utilization not captured in a single snapshot. CPU utilization data unavailable for all 55 VMs (cpu_avg_pct null fleet-wide). No EOL OS violations detected: all VMs run WS2016 (extended support until Jan 2027), WS2019, or WS2022.
This tab shows the raw output of each analysis agent — the specialized components that examine telemetry, dependencies, contracts, and cost. It is the most detailed and most technical view. Most readers will find the Overview and Backlog tabs sufficient.
Telemetry Agent
Utilization Profilemoderate
Snapshot Count1
APM AvailableFalse
Hard Rule TriggeredTrue
Right-Sizing Opp.True
Agent Confidence45%
Telemetry — VM Counts
Powered Off VMs0
Flagged VMs0
Mem Active % (avg)9.3%
Data WindowNone
Inventory Source
Dependency Agent
Epic IntegrationFalse
PHI HandledUNKNOWN
Hardware DependencyFalse
Latency SensitiveFalse
Blast Radiushigh
Migration RiskMEDIUM
Conf (Inferred)78%
Conf (Confirmed)52%
Verificationsingle_source
Cloud Connectivity
ComplianceHIPAA / HITRUST
Upstream (5)
AGFA PACS (legacy): confirmed: rvtools.vm_notes.VNAUHDMMIG03
Merge PACS (legacy): confirmed: rvtools.vm_notes.VNAMERGEMIGRATE, vnamidmerg1, vnamidmerg2
Black River source system: confirmed: rvtools.vm_notes.VNABRMIGRATE1, VNABRMIGRATE2
Orangeburg source system: confirmed: rvtools.vm_notes.VNAORGMIGRATE1, VNAORGMIGRATE2
SQL Server (migration databases): confirmed: rvtools.vm_notes.VNAUHVIEW7, vnauh2migdb, vnauhdmsql, vnauhmigdb
Downstream (4)
NilRead Diagnostic Viewers — ART datacenter: confirmed: rvtools.vm_notes.VNARHNILDMZ1, VNARHVIEW1-10, vnarhview5-7
NilRead Diagnostic Viewers — USB datacenter: confirmed: rvtools.vm_notes.VNAUHNILDMZ1, VNAUHVIEW1-10, vnauhview5-6
Hyland Research Servers: confirmed: rvtools.vm_notes.VNARHRESEAR, VNAUHRESEAR
Radiology clinical workflows (MUSC): inferred: application_name=Hyland OnBase/VNA + industry_context=healthcare + NilRead viewer cluster
Procurement Agent
SaaS Available—
SaaS VendorHyland
SaaS ProductHyland Cloud (SaaS)
Repurchase ViableINSUFFICIENT_EVIDENCE
BAA ConfirmedUNKNOWN
SaaS Confidence
Inferred Score80%
Confirmed Score0%
Verificationsingle_source
AWS Equiv.None
Provisioning Agent (Python engine — not LLM)
Current VMs55
Recommended VMs41
Right-Sizing25% reduction
Current TCO/yr$357,500
Projected Cloud TCO$213,200
Migration One-Time$42,640
Dual-Running Cost$190,233
-$88,573
Year-1 Net
$144,300
Steady-State
$200,027
3-Year
AWS TargetAmazon EC2
Azure TargetAzure Virtual Machines
Cost Modeldefault_multipliers
DEF: {'default_id': 'DEF-WL-001', 'field': 'workload_type', 'value': 'app_server', 'reason': 'workload_type indeterminable'}
DEF: {'default_id': 'DEF-MIG-001', 'field': 'migration_factor', 'value': 0.2, 'reason': 'migration_complexity not provided'}
DEF: {'default_id': 'DEF-COST-001', 'field': 'cost_model_basis', 'value': 'default_multipliers', 'reason': 'no customer cloud quote provided; using SRC-TCO-* and SRC-MULT-* registry defaults'}
CardioAccess
Vendor: Cardiocom · 1 VMs · Criticality: High · HIPAA / HITRUST — PHI handling confirmed
REHOST
55%
Confidence
28%
Data Quality
ROI: Year-1 Net: -$3,640 · Steady-State: $1,300/yr · 3-Year: -$1,040
Overview
Backlog
Provenance
Technical Details
Agents
Recommendation
REHOST ⏳ PENDING CONDITIONS
SRC-CAP-002 Binding cap · Ceiling: 55% · Threshold: 60%
Secondary: RETAIN
Confidence
55%
Viable, but specific gaps must close first.
ROI Summary (Python-computed, not LLM)
-$3,640
Year-1 Net
$1,300
Steady-State /yr
-$1,040
3-Year Total
ROI variance: ±60%
⚙ Financials computed by provisioning_engine_v3 (Python) — not LLM-generated
In plain terms: migrating costs money up front — Year 1 is a net outflow of $3,640. From Year 2 onward the app saves $1,300 every year. The 3-Year total nets these against each other.
Finding
CardioAccess is a vendor cardiology platform running on a single aging Windows Server 2012 VM with moderate utilization, making it a candidate for cloud rehosting. However, critical unknowns — whether the system integrates with physical cardiac monitoring hardware, whether it has latency constraints from modality feeds, and whether a valid Business Associate Agreement is in place — must be resolved before migration can proceed. A targeted discovery sprint covering hardware integration, BAA status, and network latency requirements is required to unlock this recommendation.
In plain terms: see the recommendation and conditions above.
Conditions to Unlock (3 items)
This recommendation becomes executable once these conditions are confirmed:
☐ Verify Business Associate Agreement with Medtronic before any cloud migration
Owner: Legal / Compliance Officer · Timeline: 10 days · Unlocks: compliance_close, cap_resolution, migration_commit
☐ Determine whether CardioAccess connects to physical cardiac devices or requires low-latency network links
Owner: Enterprise Architect / Clinical Engineering · Timeline: 10 days · Unlocks: migration_commit, cap_resolution, operational_safety
☐ Determine whether CardioAccess is classified as a life-safety system by clinical leadership
Owner: CMIO / Clinical Informatics · Timeline: 5 days · Unlocks: operational_safety, migration_commit, cap_resolution
See Backlog tab for full action backlog and P2/P3 items.
This is the list of work needed to move this recommendation forward. The 3 P1 items are the priority — they unlock the recommendation. P2 and P3 items improve confidence and reduce risk but are not blockers. 'Story points' measure effort, not time or cost — see the Methodology tab for how they work.
55%
Current
60%
Threshold
62%
Ceiling
24
Story Points
Why This Score
CardioAccess is capped at 55 by SRC-CAP-002 because the majority of evidence dimensions — dependency_mapping, procurement_contracts, architecture_diagrams, and dr_requirements — are either INFERRED or MISSING, preventing the pre-cap score of 61.75 from being realized. The ceiling equals the current score, leaving zero headroom; no backlog item can move the needle until the binding cap is resolved by converting INFERRED/MISSING dimensions to CONFIRMED through the Sprint 1 discovery actions.
Proceeding with REHOST migration planning before resolving hardware_dependency, latency_sensitive, BAA status, and life_safety_classification carries high operational and compliance risk — any one of these unknowns resolving adversely could invalidate the migration path, expose the organization to HIPAA liability, or trigger a structural RETAIN override for a High-criticality PHI-handling cardiology workload.
Action Backlog — 6 Items · 24 Story Points · Ceiling: 62%
APP-017-001 Confirm BAA status with Medtronic for CardioAccess PHI workload
P1 BAA/Legal
CardioAccess handles PHI (phi_handling=true per catalog) but baa_valid=null — no BAA confirmation exists. Cardiocom was acquired by Medtronic circa 2013; all contract and BAA inquiries must target Medtronic. Collect: (1) executed BAA document between the health system and Medtronic covering CardioAccess/CARDIOACCESS-DB, (2) confirmation that the BAA extends to any cloud-hosted deployment scenario,
+0% confidence -0% ROI variance 👤 📋 · 5 pts
APP-017-002 Resolve hardware_dependency and latency_sensitive via Cardiocom/Medtronic architecture review
P1 Architecture
hardware_dependency=UNKNOWN and latency_sensitive=UNKNOWN are the two CRITICAL intelligence gaps that determine whether REHOST is viable or whether a structural RETAIN pre-check fires. CardioAccess (Cardiocom) may integrate with physical cardiac monitoring hardware (Holter recorders, ECG carts, stress test systems) via proprietary or HL7 feeds; if hardware_dependency=true, REHOST is overridden. Re
+7% confidence -0% ROI variance 👤 📋 · 5 pts
APP-017-003 Confirm life_safety_classification with clinical stakeholders
P1 Security/Compliance
life_safety_classification=UNKNOWN for a cardiology data management platform in a production healthcare environment. If life_safety_classification=true, the structural pre-check fires and REHOST is overridden by RETAIN regardless of other evidence. CardioAccess is a data management/reporting layer (not a real-time cardiac control interface), making true unlikely but unconfirmable without clinical
+7% confidence -0% ROI variance 👤 📋 · 3 pts
APP-017-004 Engage Medtronic to determine CardioAccess SaaS/cloud-hosted roadmap
P2 Vendor Engagement
saas_alternative_available=UNKNOWN and saas_confidence_inferred=20. Cardiocom was acquired by Medtronic circa 2013; the CardioAccess product roadmap is controlled by Medtronic's cardiac remote monitoring division. A cloud-hosted or SaaS offering from Medtronic would change the strategic recommendation from REHOST to REPURCHASE if saas_confidence_inferred reaches ≥50. Collect: (1) Medtronic product
+0% confidence -0% ROI variance 👤 📋 · 3 pts
APP-017-005 Capture DR/RTO/RPO requirements for High-criticality cardiology workload
P2 Architecture
dr_requirements=MISSING for a High-criticality cardiology system. No RTO/RPO data is available to validate cloud architecture choices (multi-AZ placement, backup frequency, failover SLA). This gap contributes dr_penalty to the ROI variance calculation. Collect: (1) RTO and RPO targets from application owner and clinical leadership, (2) current backup schedule and recovery test results for CARDIOAC
+7% confidence -0% ROI variance 👤 📋 · 3 pts
APP-017-006 Deploy APM telemetry agent and collect 30-day utilization time-series for CARDIOACCESS-DB
P2 Telemetry
snapshot_count=1 and apm_telemetry_available=false. SRC-CAP-004 applies (ceiling 65) as a downstream cap once SRC-CAP-002 is resolved. A single RVTools snapshot cannot distinguish a genuinely idle VM from one sampled between burst cycles; CPU utilization is entirely absent (cpu_avg_pct=null). Right-sizing is capped at MODERATE (SRC-RS-002, 0.25 reduction) under the hard rule. Collect: (1) deploy l
+0% confidence -0% ROI variance 👤 📋 · 5 pts
This tab shows the evidence behind the recommendation. The seven dimensions are scored by quality — CONFIRMED evidence counts fully, INFERRED evidence counts less. The validator table confirms each analysis step passed its checks. Green means the step is sound.
Evidence Classification — 7 Dimensions
Dimension Classification Multiplier Contribution
Telemetry SINGLE 0.50 6.00
Dependency Mapping INFERRED 0.25 2.50
Procurement Contracts MISSING 0.00 0.00
Architecture Diagrams MISSING 0.00 0.00
Security Compliance INFERRED 0.25 1.25
Business Criticality SINGLE 0.50 2.00
Dr Requirements MISSING 0.00 0.00
Applied Cap: SRC-CAP-002
Validation Reports — Per Agent
Agent Status Provenance Counts Defaults Used Rule Failures
telemetry PASS
SRC:2CUST:7COMP:6DEF:2
 None ✓ None
dependency PASS
SRC:2CUST:2COMP:4DEF:5
 None ✓ None
procurement PASS
None ✓ None
provisioning PASS_WITH_DEFAULTS
SRC:7CUST:1COMP:9DEF:1
• {'default_id': 'DEF-WL-001', 'field': 'workload_type', 'value': 'app_server', 'r
• {'default_id': 'DEF-MIG-001', 'field': 'migration_factor', 'value': 0.2, 'reason
 ✓ None
synthesizer PASS
None ✓ None
confidence_advisor PASS
COMP:33
 None ✓ None
This section contains the full technical reasoning behind the recommendation: the rules that fired, the decision path, the confidence math, and the open intelligence gaps. It is intended for analysts and engineers. Executive readers can rely on the Overview tab.
Rules Applied
These are the specific governance rules that produced this recommendation.
▸ RULE 6 — REHOST
No other qualifying path applies — the default for virtual workloads is to rehost to cloud IaaS.
▸ SRC-CAP-002 (binding)
Confidence is capped at 55 because more than half of the seven evidence dimensions are inferred or missing — the conclusion is data-thin.
▸ SRC-CAP-003
Confidence is capped at 65 because none of the seven evidence dimensions reached CONFIRMED quality — there is no high-quality anchor for the analysis.
▸ SRC-CAP-004
Confidence is capped at 65 because telemetry is a single point-in-time snapshot with no ongoing monitoring — right-sizing needs usage patterns over time.
▸ RULE T2 — Right-Sizing Hard Cap
Right-sizing is limited to a moderate 25% reduction because telemetry is a single snapshot. More aggressive right-sizing needs 30+ days of monitoring data.
Decision Path
The step-by-step trace of how the recommendation was reached, in the system's own terms.
PRE-CHECK: hardware_dependency=UNKNOWN, latency_sensitive=UNKNOWN, life_safety_classification=UNKNOWN — all TRI-STATE UNKNOWN, pre-check does not fire. RULE 1 (REPURCHASE): saas_alternative_available=UNKNOWN, saas_confidence_inferred=20 < 50 — does not fire. RULE 2 (RETIRE): no redundancy_identified, no decommission_annotation — does not fire. RULE 3 (REFACTOR): vendor COTS (Cardiocom), not internally developed — does not fire. RULE 4 (REHOST): VM-based=true; hardware_dependency=UNKNOWN (cannot confirm absence, but also cannot confirm presence — UNKNOWN does not equal true, so structural blocker absent); latency_sensitive=UNKNOWN (same logic); REHOST selected as best-supported disposition with UNKNOWN caveats surfaced as conditions. RULE 5 (RETAIN): requires all three structural fields=true — does not fire on UNKNOWN. CONFIDENCE GATE (v4.1): final_confidence=55 < 60, strategic_recommendation=REHOST (not RETAIN) → recommendation_status=PENDING_CONDITIONS, displayed_primary=REHOST (gate does not force RETAIN in v4.1). CONDITIONS TO UNLOCK: (1) Confirm hardware_dependency true/false via Cardiocom architecture review; (2) Confirm latency_sensitive true/false for modality feed integration; (3) Execute BAA with Cardiocom or cloud host; (4) Obtain APM telemetry for second snapshot.
Cap Evaluation (all fired caps): SRC-CAP-002: ≥4 of 7 dimensions INFERRED/MISSING → ceiling 55 (ceiling=55) ← BINDING SRC-CAP-003: Zero CONFIRMED dimensions → ceiling 65 (ceiling=65) SRC-CAP-004: Single snapshot, no APM → ceiling 65 (ceiling=65)
Confidence Scoring Breakdown
How the confidence percentage was calculated, dimension by dimension.
Base score: 50 Evidence contributions: Telemetry SINGLE +6.00 Dependency Mapping INFERRED +2.50 Procurement Contracts MISSING +0.00 Architecture Diagrams MISSING +0.00 Security Compliance INFERRED +1.25 Business Criticality SINGLE +2.00 Dr Requirements MISSING +0.00 Pre-cap score: 61.75 Binding cap applied: SRC-CAP-002 Final confidence: 55%
Intelligence Gaps
Specific things we don't yet know. Each gap is a reason the confidence score is held back.
hardware_dependency — UNKNOWN
Cardiocom CardioAccess may integrate with physical cardiac monitoring hardware (Holter recorders, ECG carts, stress test systems) via proprietary or HL7 feeds. If hardware_dependency=true, the structural pre-check fires and REHOST is overridden by RETAIN. Must be confirmed before migration planning proceeds.
latency_sensitive — UNKNOWN
Real-time cardiac data acquisition may impose sub-millisecond latency requirements incompatible with cloud network round-trips. Confirmation required to validate REHOST viability.
BAA / baa_confirmed — UNKNOWN
phi_handling=true in application catalog; baa_valid=null. No BAA confirmation means PHI workload cannot be migrated to any cloud environment under HIPAA without executing a BAA with the cloud provider and confirming Cardiocom's cloud-hosted offering (if any) is BAA-covered. Blocks migration execution.
saas_alternative_available — UNKNOWN
Cardiocom may offer a cloud-hosted or SaaS version of CardioAccess. If confirmed with saas_confidence_inferred >= 50, REPURCHASE would supersede REHOST. Vendor engagement required.
APM telemetry absent — snapshot_count=1
Single RVTools snapshot with no APM agent data limits utilization confidence. SRC-CAP-004 applies. A second snapshot and APM instrumentation would raise confidence ceiling from 65 and improve right-sizing accuracy.
Windows Server 2012 EOL
CARDIOACCESS-DB runs Windows Server 2012 (64-bit), which reached end-of-life in October 2023. On-premises retention without ESU increases security and HIPAA compliance exposure. Cloud migration provides access to extended security updates on AWS/Azure/GCP/OCI.
life_safety_classification — UNKNOWN
Cardiology platforms in health systems may carry life-safety classification if they support real-time cardiac monitoring for inpatient care. If life_safety_classification=true, structural pre-check fires. Clinical stakeholder confirmation required.
DR requirements — MISSING
No DR/RTO/RPO data available for a High-criticality cardiology system. DR requirements must be captured to validate cloud architecture (multi-AZ, backup frequency, failover SLA).
EMR integration scope — UNCONFIRMED
epic_integration=false per catalog, but downstream EMR integration (non-Epic) cannot be excluded per dependency agent. Integration scope affects migration sequencing and cutover risk.
Technical Detail
A fuller technical explanation of the recommendation and its constraints.
Telemetry agent confirms a single powered-on VM (CARDIOACCESS-DB) at moderate utilization (24% memory active) with a right-sizing opportunity of 25% under SRC-RS-002, supporting cloud migration viability. Dependency agent flags UNKNOWN status on hardware_dependency, latency_sensitive, and life_safety_classification — all three are null in the catalog — which prevents the structural RETAIN pre-check from firing but also prevents Rule 4 (REHOST) from firing with full confidence; REHOST is selected as the best-supported disposition given VM-based architecture and no confirmed physical constraints. Procurement agent returns saas_confidence_inferred=20 and saas_available=UNKNOWN, ruling out REPURCHASE under Rule 1. The binding confidence cap is SRC-CAP-002 (majority dimensions INFERRED/MISSING, ceiling 55), placing final confidence at 55 and triggering PENDING_CONDITIONS status requiring discovery before execution.
Secondary Path
The fallback option if the primary recommendation cannot proceed.
Secondary Path — RETAIN
RETAIN is the secondary disposition if discovery confirms hardware_dependency=true or latency_sensitive=true due to physical cardiac modality feed integration (e.g., Holter recorders, ECG carts on proprietary protocols); in that scenario the structural pre-check would fire and override REHOST.
Right-Sizing Notes
Why the cloud sizing estimate is set where it is, based on the telemetry available.
Rule T2 (SRC-RS-002 hard cap) is active: snapshot_count=1 AND apm_telemetry_available=false. Raw fleet_avg_mem_active_pct=24.0% falls below the 25% OVERSIZED threshold under Rule T3, which would ordinarily yield SRC-RS-003 (OVERSIZED, 0.40 reduction). The single-snapshot hard rule caps the classification at SRC-RS-002 (MODERATE, 0.25 reduction) because a single point-in-time reading cannot distinguish a genuinely idle VM from one sampled between burst cycles. A right-sizing opportunity is flagged at MODERATE confidence only. To unlock a higher-fidelity tier assessment, 30+ days of APM time-series or a minimum of two RVTools snapshots separated by at least 7 days are required. CPU utilization data is entirely absent (cpu_avg_pct null on 1 of 1 VMs); this further limits confidence.
This tab shows the raw output of each analysis agent — the specialized components that examine telemetry, dependencies, contracts, and cost. It is the most detailed and most technical view. Most readers will find the Overview and Backlog tabs sufficient.
Telemetry Agent
Utilization Profilemoderate
Snapshot Count1
APM AvailableFalse
Hard Rule TriggeredTrue
Right-Sizing Opp.True
Agent Confidence45%
Telemetry — VM Counts
Powered Off VMs0
Flagged VMs0
Mem Active % (avg)24.0%
Data WindowNone
Inventory Source
Dependency Agent
Epic IntegrationFalse
PHI HandledUNKNOWN
Hardware DependencyUNKNOWN
Latency SensitiveUNKNOWN
Blast Radiushigh
Migration RiskMEDIUM
Conf (Inferred)72%
Conf (Confirmed)38%
Verificationsingle_source
Cloud Connectivity
ComplianceHIPAA / HITRUST
Upstream (2)
Cardiology data acquisition / modality feed (assumed): inferred: vendor_pattern — Cardiocom CardioAccess typically receives HL7 or proprietary feeds from cardiac monitoring…
Active Directory / Windows authentication: inferred: os=Windows Server 2012 + application_type=Client/Server — standard Windows domain authentication dependency
Downstream (2)
Cardiology clinical staff / reading physicians (end-user consumers): inferred: application_type=Client/Server + production web server note — clinical users access cardiac study results via…
Potential EMR / EHR integration (unconfirmed): inferred: industry_pattern — cardiology platforms in MUSC-class health systems commonly push results to EMR (Epic or…
Procurement Agent
SaaS Available—
SaaS VendorCardiocom
SaaS Product
Repurchase ViableUNKNOWN
BAA ConfirmedUNKNOWN
SaaS Confidence
Inferred Score20%
Confirmed Score0%
Verificationinferred_only
AWS Equiv.None
Provisioning Agent (Python engine — not LLM)
Current VMs1
Recommended VMs1
Right-Sizing25% reduction
Current TCO/yr$6,500
Projected Cloud TCO$5,200
Migration One-Time$1,040
Dual-Running Cost$3,900
-$3,640
Year-1 Net
$1,300
Steady-State
-$1,040
3-Year
AWS TargetAmazon EC2
Azure TargetAzure Virtual Machines
Cost Modeldefault_multipliers
DEF: {'default_id': 'DEF-WL-001', 'field': 'workload_type', 'value': 'app_server', 'reason': 'workload_type indeterminable'}
DEF: {'default_id': 'DEF-MIG-001', 'field': 'migration_factor', 'value': 0.2, 'reason': 'migration_complexity not provided'}
DEF: {'default_id': 'DEF-COST-001', 'field': 'cost_model_basis', 'value': 'default_multipliers', 'reason': 'no customer cloud quote provided; using SRC-TCO-* and SRC-MULT-* registry defaults'}
DynaCad Radiology
Vendor: Invivo · 1 VMs · Criticality: High · HIPAA / HITRUST
REHOST
40%
Confidence
22%
Data Quality
ROI: Year-1 Net: -$3,640 · Steady-State: $1,300/yr · 3-Year: -$1,040
Overview
Backlog
Provenance
Technical Details
Agents
Recommendation
REHOST ⏳ PENDING CONDITIONS
SRC-CAP-001 Binding cap · Ceiling: 40% · Threshold: 60%
Secondary: RETAIN
Confidence
40%
Viable, but specific gaps must close first.
ROI Summary (Python-computed, not LLM)
-$3,640
Year-1 Net
$1,300
Steady-State /yr
-$1,040
3-Year Total
ROI variance: ±60%
⚙ Financials computed by provisioning_engine_v3 (Python) — not LLM-generated
In plain terms: migrating costs money up front — Year 1 is a net outflow of $3,640. From Year 2 onward the app saves $1,300 every year. The 3-Year total nets these against each other.
Finding
DynaCad Radiology is a single-VM Windows Server workstation running at low memory utilization (9%) and is a candidate for cloud rehosting to reduce on-premises infrastructure costs. However, because this application handles patient imaging data and no Business Associate Agreement has been confirmed with the vendor, the recommendation cannot be executed until BAA status and hardware integration requirements are verified. Discovery work to confirm DICOM/PACS connectivity and any physical hardware dependencies is required before migration planning can proceed.
In plain terms: see the recommendation and conditions above.
Conditions to Unlock (2 items)
This recommendation becomes executable once these conditions are confirmed:
☐ Verify Business Associate Agreement with DynaCad Vendor Before Cloud Migration
Owner: Healthcare Compliance Officer / Legal Counsel · Timeline: 14 days · Unlocks: compliance_close, cap_resolution, migration_commit
☐ Determine Whether DynaCad Requires Physical Hardware to Operate
Owner: Infrastructure Architect / Vendor Engagement Lead · Timeline: 10 days · Unlocks: migration_commit, operational_safety
See Backlog tab for full action backlog and P2/P3 items.
This is the list of work needed to move this recommendation forward. The 2 P1 items are the priority — they unlock the recommendation. P2 and P3 items improve confidence and reduce risk but are not blockers. 'Story points' measure effort, not time or cost — see the Methodology tab for how they work.
40%
Current
60%
Threshold
55%
Ceiling
22
Story Points
Why This Score
DynaCad Radiology is capped at confidence 40 by SRC-CAP-001 (phi_handled=true with baa_confirmed=UNKNOWN), which is the binding constraint overriding the pre-cap score of 63.5; no amount of additional evidence collection can raise confidence above 40 until BAA status is resolved. Secondary caps SRC-CAP-002 (majority INFERRED/MISSING dimensions), SRC-CAP-003 (zero CONFIRMED dimensions), and SRC-CAP-004 (single snapshot, no APM) form a sequential ceiling stack that will govern confidence progression after SRC-CAP-001 is lifted.
Proceeding with cloud migration planning before BAA confirmation and hardware dependency resolution carries material compliance and technical risk — if hardware_dependency resolves to true, the entire REHOST disposition must be reversed to RETAIN, invalidating any migration work already initiated.
Action Backlog — 6 Items · 22 Story Points · Ceiling: 55%
APP-018-001 Confirm BAA with Invivo/Philips for DynaCad Cloud Hosting
P1 BAA/Legal
Retrieve or establish a Business Associate Agreement (BAA) with Invivo (Philips Healthcare) that explicitly covers DynaCad Radiology in a cloud-hosted deployment. Confirm whether the existing enterprise Philips BAA extends to DynaCad specifically, or whether a separate instrument is required. Obtain written confirmation from the compliance officer and upload to the contract repository. This is the
+15% confidence -0% ROI variance 👤 📋 · 3 pts
APP-018-002 Assess Physical Hardware Dependencies for DynaCad Radiology
P1 Dependency Mapping
Contact Invivo/Philips vendor support and review DynaCad installation documentation to determine whether the application requires physical hardware components (USB/HASP license dongles, GPU acceleration cards, direct DICOM hardware controller interfaces, or MRI scanner hardware coupling). Review VM notes and datacenter rack documentation for Dynacad5prod. If hardware_dependency=true is confirmed,
+0% confidence -0% ROI variance 👤 📋 · 3 pts
APP-018-003 Deploy 30-Day APM Telemetry Collection on Dynacad5prod
P2 Telemetry
Instrument Dynacad5prod with APM tooling (e.g., Datadog, New Relic, or native Windows Performance Monitor with 5-minute polling) to collect CPU utilization, memory active percentage, disk I/O, and network throughput over a minimum 30-day window that includes at least one full radiology workflow cycle (weekday peak, weekend low, month-end reporting). Current single-snapshot data (mem_active=9%, cpu
+0% confidence -0% ROI variance 👤 📋 · 5 pts
APP-018-004 Map DICOM/PACS Integration Topology and Latency Requirements
P2 Dependency Mapping
Conduct a structured dependency mapping exercise to identify all upstream DICOM imaging modalities and PACS systems that feed Dynacad5prod, all downstream consumers (radiologist workstations, RIS, non-Epic EMR), and network latency requirements for DICOM study retrieval. Measure current LAN latency between DynaCad and PACS. Determine whether latency_sensitive=true applies by establishing the minim
+0% confidence -0% ROI variance 👤 📋 · 5 pts
APP-018-005 Retrieve DynaCad Contract Terms and Vendor SaaS Roadmap
P2 Procurement
Retrieve the current DynaCad license agreement from the contract repository or Invivo/Philips account team to capture ACV, renewal date, and cloud deployment rights. Submit a formal vendor inquiry to Invivo/Philips asking: (1) whether DynaCad is available as a cloud-hosted or SaaS offering; (2) whether the current license permits cloud VM deployment; (3) what the vendor's cloud migration support p
+0% confidence -0% ROI variance 👤 📋 · 3 pts
APP-018-006 Document RTO/RPO Requirements for DynaCad Radiology
P3 Architecture
Conduct a DR requirements workshop with the Radiology department and IT leadership to establish RTO and RPO targets for DynaCad Radiology as a High-criticality production system. Document the acceptable downtime window for radiologist workflow continuity, backup frequency requirements for DICOM study data, and failover architecture expectations. This resolves dr_requirements=MISSING, removes the 1
+0% confidence -0% ROI variance 👤 📋 · 3 pts
This tab shows the evidence behind the recommendation. The seven dimensions are scored by quality — CONFIRMED evidence counts fully, INFERRED evidence counts less. The validator table confirms each analysis step passed its checks. Green means the step is sound.
Evidence Classification — 7 Dimensions
Dimension Classification Multiplier Contribution
Telemetry SINGLE 0.50 6.00
Dependency Mapping INFERRED 0.25 2.50
Procurement Contracts MISSING 0.00 0.00
Architecture Diagrams INFERRED 0.25 1.75
Security Compliance INFERRED 0.25 1.25
Business Criticality SINGLE 0.50 2.00
Dr Requirements MISSING 0.00 0.00
Applied Cap: SRC-CAP-001
Validation Reports — Per Agent
Agent Status Provenance Counts Defaults Used Rule Failures
telemetry PASS
SRC:2CUST:8COMP:4DEF:3
 None ✓ None
dependency PASS
SRC:6CUST:1COMP:5DEF:3
 None ✓ None
procurement PASS
None ✓ None
provisioning PASS_WITH_DEFAULTS
SRC:7CUST:1COMP:9DEF:1
• {'default_id': 'DEF-WL-001', 'field': 'workload_type', 'value': 'app_server', 'r
• {'default_id': 'DEF-MIG-001', 'field': 'migration_factor', 'value': 0.2, 'reason
 ✓ None
synthesizer PASS
None ✓ None
confidence_advisor PASS
COMP:34
 None ✓ None
This section contains the full technical reasoning behind the recommendation: the rules that fired, the decision path, the confidence math, and the open intelligence gaps. It is intended for analysts and engineers. Executive readers can rely on the Overview tab.
Rules Applied
These are the specific governance rules that produced this recommendation.
▸ RULE 6 — REHOST
No other qualifying path applies — the default for virtual workloads is to rehost to cloud IaaS.
▸ SRC-CAP-001 (binding)
Confidence is capped at 40 because the app handles PHI and the vendor's Business Associate Agreement is not confirmed. PHI cannot move to a cloud vendor without a confirmed BAA.
▸ SRC-CAP-002
Confidence is capped at 55 because more than half of the seven evidence dimensions are inferred or missing — the conclusion is data-thin.
▸ SRC-CAP-003
Confidence is capped at 65 because none of the seven evidence dimensions reached CONFIRMED quality — there is no high-quality anchor for the analysis.
▸ SRC-CAP-004
Confidence is capped at 65 because telemetry is a single point-in-time snapshot with no ongoing monitoring — right-sizing needs usage patterns over time.
▸ RULE T2 — Right-Sizing Hard Cap
Right-sizing is limited to a moderate 25% reduction because telemetry is a single snapshot. More aggressive right-sizing needs 30+ days of monitoring data.
Decision Path
The step-by-step trace of how the recommendation was reached, in the system's own terms.
PRE-CHECK: hardware_dependency=UNKNOWN, latency_sensitive=UNKNOWN, life_safety_classification=false → pre-check does not fire. RULE 1 (REPURCHASE): saas_alternative_available=UNKNOWN, saas_confidence_inferred=20 < 50 → does not fire. RULE 2 (RETIRE): no redundancy_identified, no decommission_annotation → does not fire. RULE 3 (REFACTOR): vendor COTS (Invivo), not internally developed → does not fire. RULE 4 (REHOST): VM-based=true, hardware_dependency not confirmed true, latency_sensitive not confirmed true → FIRES. strategic_recommendation=REHOST. CONFIDENCE: base=50, telemetry(SINGLE)=+6.0, dependency(INFERRED)=+2.5, procurement(MISSING)=+0.0, architecture(INFERRED)=+1.75, security(INFERRED)=+1.25, criticality(SINGLE)=+2.0, dr(MISSING)=+0.0, pre-cap=63.5. CAPS: SRC-CAP-001(phi+no BAA)=40, SRC-CAP-002(majority INFERRED/MISSING)=55, SRC-CAP-003(zero CONFIRMED)=65, SRC-CAP-004(snapshot=1+no APM)=65. Binding cap=SRC-CAP-001, final_confidence=40. GATE (v4.1): final_confidence=40 < 60, strategic_recommendation=REHOST (not RETAIN) → recommendation_status=PENDING_CONDITIONS, displayed_primary=REHOST, confidence_gate_applied=true.
Cap Evaluation (all fired caps): SRC-CAP-001: PHI handled + BAA not confirmed → ceiling 40 (ceiling=40) ← BINDING SRC-CAP-002: ≥4 of 7 dimensions INFERRED/MISSING → ceiling 55 (ceiling=55) SRC-CAP-003: Zero CONFIRMED dimensions → ceiling 65 (ceiling=65) SRC-CAP-004: Single snapshot, no APM → ceiling 65 (ceiling=65)
Confidence Scoring Breakdown
How the confidence percentage was calculated, dimension by dimension.
Base score: 50 Evidence contributions: Telemetry SINGLE +6.00 Dependency Mapping INFERRED +2.50 Procurement Contracts MISSING +0.00 Architecture Diagrams INFERRED +1.75 Security Compliance INFERRED +1.25 Business Criticality SINGLE +2.00 Dr Requirements MISSING +0.00 Pre-cap score: 63.5 Binding cap applied: SRC-CAP-001 Final confidence: 40%
Intelligence Gaps
Specific things we don't yet know. Each gap is a reason the confidence score is held back.
BAA Confirmation Required (P1 — Binding Cap)
phi_handled=true with baa_confirmed=UNKNOWN triggers SRC-CAP-001, capping confidence at 40. BAA must be confirmed with Invivo before any cloud migration can proceed. This is the single binding condition to unlock EXECUTABLE status.
Hardware Dependency Unresolved
hardware_dependency=UNKNOWN. DynaCad is a Philips/Invivo MRI post-processing workstation that may require hardware dongles, GPU acceleration, or direct DICOM hardware interfaces. If hardware_dependency=true is confirmed, the structural pre-check fires and strategic_recommendation changes to RETAIN.
DICOM / PACS Integration Topology Unconfirmed
All upstream dependencies (DICOM Imaging Modalities, PACS) are inferred only. Network latency requirements for DICOM study retrieval in a cloud environment are unknown. If latency_sensitive=true is confirmed, the structural pre-check fires.
No APM Telemetry Available
apm_telemetry_available=false with snapshot_count=1. Application-layer performance, peak utilization patterns, and DICOM throughput requirements cannot be assessed. Right-sizing recommendation (SRC-RS-002, 25% reduction) carries elevated uncertainty.
Procurement / Contract Intelligence Missing
contract_intelligence=INSUFFICIENT_EVIDENCE. License portability to cloud, vendor support for cloud deployment, and SaaS roadmap for DynaCad are all unknown. Procurement penalty of 15% applied to ROI variance.
DR Requirements Undocumented
No DR documentation available for a High-criticality radiology workstation. RTO/RPO targets for radiologist workflow continuity are unknown. DR penalty of 10% applied to ROI variance.
Technical Detail
A fuller technical explanation of the recommendation and its constraints.
Telemetry confirms a single powered-on VM (Dynacad5prod) running Windows Server 2019 at 9% memory utilization with a right-sizing opportunity of 25% (SRC-RS-002). The dependency agent identifies hardware_dependency=UNKNOWN and latency_sensitive=UNKNOWN, meaning the structural retain pre-check does not fire; Rule 4 (REHOST) fires as the first matching rule for a VM-based workload with no confirmed constraints. The procurement agent returns saas_confidence_inferred=20, which is below the Rule 1 threshold of 50, so REPURCHASE is not viable. SRC-CAP-001 applies as the binding cap (phi_handled=true, baa_confirmed=UNKNOWN), reducing final confidence to 40 and triggering PENDING_CONDITIONS status; the recommendation remains REHOST per v4.1 gate behavior.
Secondary Path
The fallback option if the primary recommendation cannot proceed.
Secondary Path — RETAIN
RETAIN is the secondary disposition if discovery confirms hardware_dependency=true or latency_sensitive=true for DICOM real-time processing requirements, which would trigger the structural pre-check.
Right-Sizing Notes
Why the cloud sizing estimate is set where it is, based on the telemetry available.
RULE T2 HARD RULE TRIGGERED: snapshot_count=1 AND apm_telemetry_available=false. Fleet avg mem_active_pct is 9.0%, which under RULE T3 would qualify for SRC-RS-003 (OVERSIZED, <25%). However, single-snapshot data cannot distinguish a genuinely idle VM from one sampled between burst cycles. Per SRC-RS-002 hard cap, tier is capped at MODERATE (0.25 reduction). ZOMBIE and OVERSIZED tiers are FORBIDDEN for this run. RULE T1 did not fire: application_type='Client/Server' (not 'Hardware-Integrated') and life_safety_classification is absent from catalog. CPU utilization data is entirely absent (cpu_avg_pct null on 1 of 1 VMs); no APM time-series provided. OS (Windows Server 2019) is within extended support through January 2029; no EOL compliance flags generated. phi_handling=true and baa_valid=null are surfaced for synthesizer consumption.
This tab shows the raw output of each analysis agent — the specialized components that examine telemetry, dependencies, contracts, and cost. It is the most detailed and most technical view. Most readers will find the Overview and Backlog tabs sufficient.
Telemetry Agent
Utilization Profilemoderate
Snapshot Count1
APM AvailableFalse
Hard Rule TriggeredTrue
Right-Sizing Opp.True
Agent Confidence45%
Telemetry — VM Counts
Powered Off VMs0
Flagged VMs0
Mem Active % (avg)9.0%
Data WindowNone
Inventory Source
Dependency Agent
Epic IntegrationFalse
PHI HandledTrue
Hardware DependencyUNKNOWN
Latency SensitiveUNKNOWN
Blast Radiusmedium
Migration RiskMEDIUM
Conf (Inferred)62%
Conf (Confirmed)38%
Verificationsingle_source
Cloud Connectivity
ComplianceHIPAA / HITRUST
Upstream (2)
DICOM Imaging Modalities / PACS: inferred: application_name='DynaCad Radiology' + vendor='Invivo' (Philips MRI workstation product line)
Active Directory / Windows Authentication: inferred: os='Windows Server 2019' + application_type='Client/Server'
Downstream (2)
Radiologist Workstations / Clinical Review Clients: inferred: application_type='Client/Server' implies client endpoints consume server output
RIS / EMR (potential): inferred: radiology workflow pattern in healthcare; epic_integration=false rules out Epic specifically but not other…
Procurement Agent
SaaS Available—
SaaS VendorInvivo
SaaS Product
Repurchase ViableUNKNOWN
BAA ConfirmedUNKNOWN
SaaS Confidence
Inferred Score20%
Confirmed Score0%
Verificationinferred_only
AWS Equiv.None
Provisioning Agent (Python engine — not LLM)
Current VMs1
Recommended VMs1
Right-Sizing25% reduction
Current TCO/yr$6,500
Projected Cloud TCO$5,200
Migration One-Time$1,040
Dual-Running Cost$3,900
-$3,640
Year-1 Net
$1,300
Steady-State
-$1,040
3-Year
AWS TargetAmazon EC2
Azure TargetAzure Virtual Machines
Cost Modeldefault_multipliers
DEF: {'default_id': 'DEF-WL-001', 'field': 'workload_type', 'value': 'app_server', 'reason': 'workload_type indeterminable'}
DEF: {'default_id': 'DEF-MIG-001', 'field': 'migration_factor', 'value': 0.2, 'reason': 'migration_complexity not provided'}
DEF: {'default_id': 'DEF-COST-001', 'field': 'cost_model_basis', 'value': 'default_multipliers', 'reason': 'no customer cloud quote provided; using SRC-TCO-* and SRC-MULT-* registry defaults'}
Medicom
Vendor: Medicom · 1 VMs · Criticality: High · HIPAA / HITRUST
REHOST
55%
Confidence
28%
Data Quality
ROI: Year-1 Net: -$3,640 · Steady-State: $1,300/yr · 3-Year: -$1,040
Overview
Backlog
Provenance
Technical Details
Agents
Recommendation
REHOST ⏳ PENDING CONDITIONS
SRC-CAP-002 Binding cap · Ceiling: 55% · Threshold: 60%
Secondary: RETAIN (pending discovery resolution)
Confidence
55%
Viable, but specific gaps must close first.
ROI Summary (Python-computed, not LLM)
-$3,640
Year-1 Net
$1,300
Steady-State /yr
-$1,040
3-Year Total
ROI variance: ±55%
⚙ Financials computed by provisioning_engine_v3 (Python) — not LLM-generated
In plain terms: migrating costs money up front — Year 1 is a net outflow of $3,640. From Year 2 onward the app saves $1,300 every year. The 3-Year total nets these against each other.
Finding
Medicom is a single-VM radiology informatics platform at MUSC-USB with moderate utilization and no confirmed hardware or life-safety constraints that would block cloud migration. A standard rehost to IaaS is the recommended path, projecting modest annual savings of approximately $1,300. However, critical gaps in procurement contracts, BAA status, and dependency mapping must be resolved before migration can be executed safely.
In plain terms: see the recommendation and conditions above.
Conditions to Unlock (3 items)
This recommendation becomes executable once these conditions are confirmed:
☐ Verify Business Associate Agreement and Patient Data Handling Before Any Cloud Move
Owner: Compliance Officer / Legal Counsel · Timeline: 21 days · Unlocks: compliance_close, cap_resolution, migration_commit
☐ Confirm Whether Medicom Connects to Physical Radiology Hardware That Would Block Cloud Migration
Owner: Application Owner (Wendy Ketchum, Radiology Informatics) + Infrastructure Engineer · Timeline: 14 days · Unlocks: migration_commit, cap_resolution, operational_safety
☐ Obtain Medicom Contract Details and Ask Vendor About Cloud or SaaS Options
Owner: Procurement / Vendor Management · Timeline: 30 days · Unlocks: cost_clarity, cap_resolution, migration_commit
See Backlog tab for full action backlog and P2/P3 items.
This is the list of work needed to move this recommendation forward. The 3 P1 items are the priority — they unlock the recommendation. P2 and P3 items improve confidence and reduce risk but are not blockers. 'Story points' measure effort, not time or cost — see the Methodology tab for how they work.
55%
Current
60%
Threshold
62%
Ceiling
22
Story Points
Why This Score
Medicom's confidence is capped at 55 by SRC-CAP-002 because 5 of 7 evidence dimensions are INFERRED or MISSING — procurement_contracts, architecture_diagrams, and dr_requirements are entirely absent, while dependency_mapping and security_compliance rest on vendor-class inference only. The pre-cap score of 61.75 would clear the 60-point threshold, but no dimension has reached CONFIRMED status, making the cap binding until the Sprint 1 and Sprint 2 discovery items are resolved.
Executing a cloud migration of this PHI-likely radiology informatics system before BAA confirmation, hardware dependency verification, and procurement contract review creates material HIPAA compliance exposure and risks invalidating the REHOST recommendation entirely if hardware coupling is confirmed.
Action Backlog — 6 Items · 22 Story Points · Ceiling: 62%
APP-019-001 Confirm BAA Status and PHI Scope with Medicom Vendor and Target Cloud Provider
P1 BAA/Legal
catalog.baa_valid=null and phi_handling=true. The dependency agent assessed phi_risk as likely_phi based on radiology informatics context. Any cloud migration of a PHI-handling system without a confirmed BAA is a HIPAA violation. Collect: (1) executed BAA with Medicom vendor confirming cloud deployment rights; (2) BAA with target cloud provider (AWS, Azure, GCP, or OCI); (3) written confirmation o
+0% confidence -0% ROI variance 👤 📋 · 3 pts
APP-019-002 Verify Hardware Dependency via On-Site Assessment and Vendor Documentation
P1 Dependency Mapping
hardware_dependency=UNKNOWN. Medicom is a radiology informatics platform that may integrate directly with DICOM-capable modalities, PACS hardware, or radiology-specific network appliances. If hardware_dependency is confirmed true, the structural RETAIN pre-check fires and REHOST is invalidated. Collect: (1) application owner interview with Wendy Ketchum (Radiology Informatics) to enumerate physica
+0% confidence -0% ROI variance 👤 📋 · 3 pts
APP-019-003 Collect Medicom Procurement Contract, License Terms, and Vendor SaaS Roadmap
P1 Procurement
procurement_contracts=MISSING and contract_intelligence=INSUFFICIENT_EVIDENCE. No ACV, renewal date, license portability terms, or cloud deployment rights are available. saas_confidence_inferred=20 — insufficient to evaluate REPURCHASE. Collect: (1) executed Medicom license/maintenance contract with ACV and renewal date; (2) license portability clause confirming cloud VM deployment is permitted; (
+7% confidence -0% ROI variance 👤 📋 · 5 pts
APP-019-004 Map Upstream and Downstream Dependencies via Network Scan and Application Owner Interview
P2 Dependency Mapping
dependency_mapping=INFERRED. Upstream (RIS/HIS/Epic Radiology) and downstream (PACS viewers, reporting systems, clinical portals) connections are inferred from vendor-class patterns only. Blast radius is medium with unconfirmed downstream consumers. Collect: (1) network flow analysis or firewall log review for medicomvaprd identifying all inbound/outbound connections; (2) application owner intervi
+0% confidence -0% ROI variance 👤 📋 · 5 pts
APP-019-005 Document DR Requirements: RTO, RPO, and Backup Architecture for Medicom
P2 Architecture
dr_requirements=MISSING. Medicom is classified High criticality with no documented RTO/RPO. A high-criticality radiology informatics system cannot be safely migrated without established recovery objectives and a validated cloud DR architecture. Collect: (1) business owner sign-off on RTO and RPO targets; (2) current backup schedule and retention policy for medicomvaprd; (3) cloud DR architecture r
+7% confidence -0% ROI variance 👤 📋 · 3 pts
APP-019-006 Collect Architecture Diagrams Showing Medicom Integration Topology
P3 Architecture
architecture_diagrams=MISSING. No architecture artifacts are available. Integration topology with upstream RIS/HIS/Epic Radiology and downstream PACS is unknown, preventing accurate migration scoping and cutover planning. Collect: (1) existing architecture or network diagrams from Medicom vendor or internal IT; (2) if no diagrams exist, conduct a discovery workshop with the application owner to pr
+7% confidence -0% ROI variance 👤 📋 · 3 pts
This tab shows the evidence behind the recommendation. The seven dimensions are scored by quality — CONFIRMED evidence counts fully, INFERRED evidence counts less. The validator table confirms each analysis step passed its checks. Green means the step is sound.
Evidence Classification — 7 Dimensions
Dimension Classification Multiplier Contribution
Telemetry SINGLE 0.50 6.00
Dependency Mapping INFERRED 0.25 2.50
Procurement Contracts MISSING 0.00 0.00
Architecture Diagrams MISSING 0.00 0.00
Security Compliance INFERRED 0.25 1.25
Business Criticality SINGLE 0.50 2.00
Dr Requirements MISSING 0.00 0.00
Applied Cap: SRC-CAP-002
Validation Reports — Per Agent
Agent Status Provenance Counts Defaults Used Rule Failures
telemetry PASS
SRC:2CUST:9COMP:4DEF:2
 None ✓ None
dependency PASS
SRC:2CUST:2COMP:4DEF:5
 None ✓ None
procurement PASS
None ✓ None
provisioning PASS_WITH_DEFAULTS
SRC:7CUST:1COMP:9DEF:1
• {'default_id': 'DEF-WL-001', 'field': 'workload_type', 'value': 'app_server', 'r
• {'default_id': 'DEF-MIG-001', 'field': 'migration_factor', 'value': 0.2, 'reason
 ✓ None
synthesizer PASS
None ✓ None
confidence_advisor PASS
COMP:27
 None ✓ None
This section contains the full technical reasoning behind the recommendation: the rules that fired, the decision path, the confidence math, and the open intelligence gaps. It is intended for analysts and engineers. Executive readers can rely on the Overview tab.
Rules Applied
These are the specific governance rules that produced this recommendation.
▸ RULE 6 — REHOST
No other qualifying path applies — the default for virtual workloads is to rehost to cloud IaaS.
▸ SRC-CAP-002 (binding)
Confidence is capped at 55 because more than half of the seven evidence dimensions are inferred or missing — the conclusion is data-thin.
▸ SRC-CAP-003
Confidence is capped at 65 because none of the seven evidence dimensions reached CONFIRMED quality — there is no high-quality anchor for the analysis.
▸ SRC-CAP-004
Confidence is capped at 65 because telemetry is a single point-in-time snapshot with no ongoing monitoring — right-sizing needs usage patterns over time.
▸ RULE T2 — Right-Sizing Hard Cap
Right-sizing is limited to a moderate 25% reduction because telemetry is a single snapshot. More aggressive right-sizing needs 30+ days of monitoring data.
Decision Path
The step-by-step trace of how the recommendation was reached, in the system's own terms.
PRE-CHECK: hardware_dependency=UNKNOWN, latency_sensitive=UNKNOWN, life_safety_classification=UNKNOWN — all tri-state UNKNOWN, pre-check does not fire. RULE 1 (REPURCHASE): saas_alternative_available=UNKNOWN (not true), saas_confidence_inferred=20 (<50) — does not fire. RULE 2 (RETIRE): no redundancy_identified, no decommission_annotation_present — does not fire. RULE 3 (REFACTOR): vendor COTS (not internally developed) — does not fire. RULE 4 (REHOST): VM-based=true, hardware_dependency=UNKNOWN (not confirmed true), latency_sensitive=UNKNOWN (not confirmed true) — FIRES. Strategic recommendation = REHOST. CONFIDENCE: base=50, telemetry(SINGLE)=+6.0, dependency_mapping(INFERRED)=+2.5, procurement_contracts(MISSING)=+0.0, architecture_diagrams(MISSING)=+0.0, security_compliance(INFERRED)=+1.25, business_criticality(SINGLE)=+2.0, dr_requirements(MISSING)=+0.0 → pre-cap=61.75. CAP EVALUATION: SRC-CAP-001 does not apply (phi_handled=UNKNOWN, not confirmed true). SRC-CAP-002 applies (5/7 dimensions INFERRED or MISSING → majority) → cap 55. SRC-CAP-003 applies (zero CONFIRMED) → cap 65. SRC-CAP-004 applies (snapshot_count=1, apm_telemetry_available=false) → cap 65. Lowest cap = SRC-CAP-002 at 55. Final confidence = 55. GATE: 55 < 60 AND strategic_recommendation=REHOST (not RETAIN) → recommendation_status=PENDING_CONDITIONS, displayed_primary=REHOST (gate does not force RETAIN in v4.1), confidence_gate_applied=true.
Cap Evaluation (all fired caps): SRC-CAP-002: ≥4 of 7 dimensions INFERRED/MISSING → ceiling 55 (ceiling=55) ← BINDING SRC-CAP-003: Zero CONFIRMED dimensions → ceiling 65 (ceiling=65) SRC-CAP-004: Single snapshot, no APM → ceiling 65 (ceiling=65)
Confidence Scoring Breakdown
How the confidence percentage was calculated, dimension by dimension.
Base score: 50 Evidence contributions: Telemetry SINGLE +6.00 Dependency Mapping INFERRED +2.50 Procurement Contracts MISSING +0.00 Architecture Diagrams MISSING +0.00 Security Compliance INFERRED +1.25 Business Criticality SINGLE +2.00 Dr Requirements MISSING +0.00 Pre-cap score: 61.75 Binding cap applied: SRC-CAP-002 Final confidence: 55%
Intelligence Gaps
Specific things we don't yet know. Each gap is a reason the confidence score is held back.
BAA / PHI Handling Confirmation
Catalog flags phi_handling=true but dependency agent resolved phi_handled=UNKNOWN and baa_valid=null. Any cloud migration of a PHI-handling system without a confirmed BAA is a HIPAA violation. Must confirm BAA with target cloud provider and validate PHI scope before migration execution.
Hardware Dependency Verification
Medicom as a radiology informatics platform may integrate directly with DICOM-capable modalities, PACS hardware, or radiology-specific network appliances. If hardware_dependency is confirmed true, the structural RETAIN pre-check fires and REHOST is invalidated. Requires on-site or vendor confirmation.
Procurement Contract Intelligence
No contract data available (INSUFFICIENT_EVIDENCE). Cannot assess license portability, cloud deployment rights, vendor SaaS roadmap, or migration cost obligations. Procurement discovery required before any migration commitment.
Downstream Dependency Mapping
Downstream consumers (PACS viewers, reporting systems, clinical portals) are inferred from vendor-class patterns only. Unconfirmed blast radius of medium. A cutover without confirmed downstream dependency mapping risks radiology workflow disruption.
DR / RTO-RPO Requirements
No DR requirements data available. High-criticality radiology informatics system with no documented RTO/RPO cannot be safely migrated without establishing recovery objectives and validating cloud DR architecture.
Architecture Diagrams
No architecture artifacts available. Integration topology with upstream RIS/HIS/Epic Radiology and downstream PACS is unknown. Required for accurate migration scoping and cutover planning.
SaaS Alternative Assessment
saas_confidence_inferred=20 — insufficient to evaluate REPURCHASE. If Medicom or a competing vendor offers a cloud-native SaaS radiology informatics platform, REPURCHASE may be preferable to REHOST. Requires vendor engagement.
Technical Detail
A fuller technical explanation of the recommendation and its constraints.
Telemetry agent confirms a single powered-on VM (medicomvaprd) with 5% average memory utilization, triggering SRC-RS-002 right-sizing at 25% reduction — indicating MODERATE utilization and no performance pressure. Dependency agent found no confirmed hardware dependency, latency sensitivity, or life-safety classification (all UNKNOWN), meaning the structural RETAIN pre-check does not fire. Procurement agent returned UNKNOWN for SaaS availability and BAA confirmation with saas_confidence_inferred=20, insufficient to trigger Rule 1 (REPURCHASE requires ≥50). No decommission annotation or named successor was found, ruling out Rule 2 (RETIRE). Medicom is vendor COTS, ruling out Rule 3 (REFACTOR). Rule 4 (REHOST) fires: VM-based, no confirmed hardware dependency, no confirmed latency constraint.
Secondary Path
The fallback option if the primary recommendation cannot proceed.
Secondary Path — RETAIN (pending discovery resolution)
If discovery confirms a SaaS-based radiology informatics offering from Medicom or a comparable vendor with saas_confidence_inferred ≥ 50, Rule 1 (REPURCHASE) should be re-evaluated. If hardware integration with radiology modalities (e.g., DICOM gateway, direct PACS hardware) is confirmed, the structural RETAIN pre-check must be re-run.
Right-Sizing Notes
Why the cloud sizing estimate is set where it is, based on the telemetry available.
Single point-in-time RVTools snapshot with no APM telemetry. RULE T2 (SRC-RS-002 hard cap) fires: snapshot_count=1 AND apm_telemetry_available=false. Observed fleet_avg_mem_active_pct=5.0% would suggest SRC-RS-003 (OVERSIZED) under RULE T3, but that tier is FORBIDDEN under T2. Classification is capped at SRC-RS-002 (MODERATE, 0.25 reduction). A single snapshot cannot distinguish a genuinely idle VM from one sampled between burst cycles. RULE T1 does not apply: application_type=Client/Server (not Hardware-Integrated) and life_safety_classification is absent from catalog. CPU utilization data is entirely absent from the RVTools export (cpu_avg_pct=null on 1 of 1 VMs); no APM supplement available. A TELEMETRY-DEPTH-INSUFFICIENT flag has been raised to surface this gap. phi_handling=true with baa_valid=null (UNKNOWN) triggers SRC-CAP-001 confidence ceiling of 40; combined with SRC-CAP-004 (snapshot_count=1, apm_telemetry_available=false, ceiling 65), the binding cap is 40. Agent confidence set to 35 reflecting single-snapshot depth, absent CPU telemetry, and unresolved BAA status.
This tab shows the raw output of each analysis agent — the specialized components that examine telemetry, dependencies, contracts, and cost. It is the most detailed and most technical view. Most readers will find the Overview and Backlog tabs sufficient.
Telemetry Agent
Utilization Profilemoderate
Snapshot Count1
APM AvailableFalse
Hard Rule TriggeredTrue
Right-Sizing Opp.True
Agent Confidence35%
Telemetry — VM Counts
Powered Off VMs0
Flagged VMs0
Mem Active % (avg)5.0%
Data WindowNone
Inventory Source
Dependency Agent
Epic IntegrationFalse
PHI HandledUNKNOWN
Hardware DependencyUNKNOWN
Latency SensitiveUNKNOWN
Blast Radiusmedium
Migration RiskMEDIUM
Conf (Inferred)62%
Conf (Confirmed)28%
Verificationsingle_source
Cloud Connectivity
ComplianceHIPAA / HITRUST
Upstream (1)
Radiology Informatics Workflow (MUSC-USB): inferred: vm_notes reference to Radiology Informatics department implies upstream ordering/scheduling systems (e.g.,…
Downstream (1)
Radiology Informatics Consumers (unspecified): inferred: Medicom as a radiology informatics platform typically feeds downstream PACS viewers, reporting systems, or…
Procurement Agent
SaaS Available—
SaaS Vendor
SaaS Product
Repurchase ViableUNKNOWN
BAA ConfirmedUNKNOWN
SaaS Confidence
Inferred Score20%
Confirmed Score0%
Verificationinferred_only
AWS Equiv.None
Provisioning Agent (Python engine — not LLM)
Current VMs1
Recommended VMs1
Right-Sizing25% reduction
Current TCO/yr$6,500
Projected Cloud TCO$5,200
Migration One-Time$1,040
Dual-Running Cost$3,900
-$3,640
Year-1 Net
$1,300
Steady-State
-$1,040
3-Year
AWS TargetAmazon EC2
Azure TargetAzure Virtual Machines
Cost Modeldefault_multipliers
DEF: {'default_id': 'DEF-WL-001', 'field': 'workload_type', 'value': 'app_server', 'reason': 'workload_type indeterminable'}
DEF: {'default_id': 'DEF-MIG-001', 'field': 'migration_factor', 'value': 0.2, 'reason': 'migration_complexity not provided'}
DEF: {'default_id': 'DEF-COST-001', 'field': 'cost_model_basis', 'value': 'default_multipliers', 'reason': 'no customer cloud quote provided; using SRC-TCO-* and SRC-MULT-* registry defaults'}
Infinitt Dental PACS
Vendor: Infinitt · 12 VMs · Criticality: High · HIPAA / HITRUST — PHI handled; BAA requi
REPURCHASE
40%
Confidence
28%
Data Quality
ROI: Year-1 Net: -$19,760 · Steady-State: $31,200/yr · 3-Year: $42,640
Overview
Backlog
Provenance
Technical Details
Agents
Recommendation
REPURCHASE ⏳ PENDING CONDITIONS
SRC-CAP-001 Binding cap · Ceiling: 40% · Threshold: 60%
Secondary: REHOST
Confidence
40%
Viable, but specific gaps must close first.
ROI Summary (Python-computed, not LLM)
-$19,760
Year-1 Net
$31,200
Steady-State /yr
$42,640
3-Year Total
ROI variance: ±60%
⚙ Financials computed by provisioning_engine_v3 (Python) — not LLM-generated
In plain terms: migrating costs money up front — Year 1 is a net outflow of $19,760. From Year 2 onward the app saves $31,200 every year. The 3-Year total nets these against each other.
Finding
Infinitt Dental PACS is a strong candidate for migration to Infinitt Cloud PACS, the same vendor's hosted offering, which would eliminate all 12 on-premises VMs and reduce annual infrastructure costs by approximately $31,200. However, because this system handles protected health information and no Business Associate Agreement has been confirmed with the cloud vendor, the recommendation cannot be executed until BAA status is resolved and contract terms are validated. Once those conditions are met, this is a straightforward same-vendor SaaS transition with a clear migration path.
In plain terms: this app should move to the vendor's cloud product, but a few specific items must be confirmed first. See Conditions to Unlock below.
Conditions to Unlock (3 items)
This recommendation becomes executable once these conditions are confirmed:
☐ Obtain Business Associate Agreement with Infinitt for Cloud PACS
Owner: Compliance Officer / Legal Counsel · Timeline: 21 days · Unlocks: cap_resolution, compliance_close, migration_commit
☐ Request and Review Infinitt Cloud PACS Contract and Pricing Proposal
Owner: Procurement / Vendor Management · Timeline: 21 days · Unlocks: cost_clarity, migration_commit
☐ Map All System Connections for Infinitt Dental PACS
Owner: Enterprise Architect / Integration Team · Timeline: 14 days · Unlocks: migration_commit, operational_safety
See Backlog tab for full action backlog and P2/P3 items.
This is the list of work needed to move this recommendation forward. The 3 P1 items are the priority — they unlock the recommendation. P2 and P3 items improve confidence and reduce risk but are not blockers. 'Story points' measure effort, not time or cost — see the Methodology tab for how they work.
40%
Current
60%
Threshold
55%
Ceiling
32
Story Points
Why This Score
Final confidence is 40, hard-capped by SRC-CAP-001 (phi_handled=true AND baa_confirmed=UNKNOWN), which overrides the pre-cap score of 63.5 and is the lowest of four simultaneously active caps (SRC-CAP-001 at 40, SRC-CAP-002 at 55, SRC-CAP-003 at 65, SRC-CAP-004 at 65). No backlog item can raise confidence in this run because current_confidence equals the ceiling; the entire backlog is oriented toward resolving the binding cap and reducing the 60% ROI variance that results from five MISSING or INFERRED evidence dimensions.
Executing REPURCHASE to Infinitt Cloud PACS without a confirmed BAA constitutes a HIPAA violation risk for PHI in a vendor-hosted environment; no migration action should proceed until APP-020-001 (BAA confirmation) is closed.
Action Backlog — 6 Items · 32 Story Points · Ceiling: 55%
APP-020-001 Confirm BAA with Infinitt for Cloud PACS — resolve SRC-CAP-001 binding cap
P1 BAA/Legal
Retrieve or execute a signed Business Associate Agreement (BAA) with Infinitt Healthcare covering the Cloud PACS vendor-hosted offering. Confirm the BAA explicitly names the entity (MUSC dental operations), covers PHI stored and processed in the cloud environment, and includes breach notification timelines compliant with HIPAA §164.314. Without a confirmed BAA, phi_handled=true AND baa_confirmed !
+15% confidence -0% ROI variance 👤 📋 · 5 pts
APP-020-002 Obtain Infinitt Cloud PACS contract terms — pricing, SLA, data migration, and exit provisions
P1 Procurement
Engage Infinitt Healthcare to obtain a customer-specific RFP response or migration proposal for Cloud PACS. Required artifacts: (1) ACV/subscription pricing for the institution's imaging volume and user count; (2) data migration SLA including DICOM archive transfer timeline and validation protocol; (3) uptime SLA and support tier; (4) exit/data portability provisions. Currently contract_intelligen
+0% confidence -0% ROI variance 👤 📋 · 5 pts
APP-020-003 Conduct formal dependency mapping — HL7 feeds, DICOM modalities, dose consumers, network gateway
P1 Dependency Mapping
Execute a formal dependency mapping exercise to confirm all upstream and downstream integrations currently inferred from VM naming patterns. Required confirmations: (1) HL7 feed source system identity and message types (ADT, ORM, ORU) flowing through CDMINFHL701/CDMINFHL7T01; (2) specific dental imaging acquisition devices (X-ray units, panoramic sensors, CBCT scanners) sending DICOM to the PACS —
+0% confidence -0% ROI variance 👤 📋 · 8 pts
APP-020-004 Deploy APM telemetry — 30-day application-layer performance baseline for DICOM throughput and user load
P2 Telemetry
Deploy application performance monitoring (APM) across the Infinitt Dental PACS fleet to collect 30+ days of time-series CPU, RAM, DICOM throughput, concurrent user sessions, and image retrieval latency. Currently apm_telemetry_available=false and snapshot_count=1, triggering SRC-CAP-004 (ceiling 65) and hard rule T2 (right-sizing capped at MODERATE/25%). APM data will: (1) validate or revise the
+0% confidence -0% ROI variance 👤 📋 · 8 pts
APP-020-005 Document DR/RTO-RPO requirements for High-criticality PHI imaging system
P2 Architecture
Obtain or produce formal DR documentation for Infinitt Dental PACS specifying RTO (Recovery Time Objective) and RPO (Recovery Point Objective) requirements. Currently dr_requirements=MISSING, contributing 0/4 weight points and adding a dr_penalty to ROI variance. For a High-criticality clinical imaging system handling PHI, RTO/RPO requirements govern: (1) cloud architecture selection (active-activ
+0% confidence -0% ROI variance 👤 📋 · 3 pts
APP-020-006 Confirm latency sensitivity and DICOM modality hardware gateway dependencies
P2 Architecture
Resolve two related UNKNOWN fields that affect SaaS viability: (1) latency_sensitive — interview clinical operations and dental radiology staff to determine whether sub-second DICOM image retrieval is a contractual, regulatory, or workflow requirement. If confirmed true, evaluate whether Infinitt Cloud PACS SLA and dedicated connectivity (Direct Connect / ExpressRoute) satisfy the requirement. (2)
+0% confidence -0% ROI variance 👤 📋 · 3 pts
This tab shows the evidence behind the recommendation. The seven dimensions are scored by quality — CONFIRMED evidence counts fully, INFERRED evidence counts less. The validator table confirms each analysis step passed its checks. Green means the step is sound.
Evidence Classification — 7 Dimensions
Dimension Classification Multiplier Contribution
Telemetry SINGLE 0.50 6.00
Dependency Mapping INFERRED 0.25 2.50
Procurement Contracts MISSING 0.00 0.00
Architecture Diagrams INFERRED 0.25 1.75
Security Compliance INFERRED 0.25 1.25
Business Criticality SINGLE 0.50 2.00
Dr Requirements MISSING 0.00 0.00
Applied Cap: SRC-CAP-001
Validation Reports — Per Agent
Agent Status Provenance Counts Defaults Used Rule Failures
telemetry PASS
SRC:2CUST:10COMP:4DEF:2
 None ✓ None
dependency PASS
SRC:9COMP:6DEF:3
 None ✓ None
procurement PASS
None ✓ None
provisioning PASS_WITH_DEFAULTS
SRC:7CUST:1COMP:9DEF:1
• {'default_id': 'DEF-WL-001', 'field': 'workload_type', 'value': 'app_server', 'r
• {'default_id': 'DEF-MIG-001', 'field': 'migration_factor', 'value': 0.2, 'reason
 ✓ None
synthesizer PASS
None ✓ None
confidence_advisor PASS
COMP:15
 None ✓ None
This section contains the full technical reasoning behind the recommendation: the rules that fired, the decision path, the confidence math, and the open intelligence gaps. It is intended for analysts and engineers. Executive readers can rely on the Overview tab.
Rules Applied
These are the specific governance rules that produced this recommendation.
▸ RULE 1 — REPURCHASE
A vendor SaaS replacement is available and credible, and the app has no hardware constraint — so the recommendation is to replace it with the vendor's cloud service.
▸ SRC-CAP-001 (binding)
Confidence is capped at 40 because the app handles PHI and the vendor's Business Associate Agreement is not confirmed. PHI cannot move to a cloud vendor without a confirmed BAA.
▸ SRC-CAP-002
Confidence is capped at 55 because more than half of the seven evidence dimensions are inferred or missing — the conclusion is data-thin.
▸ SRC-CAP-003
Confidence is capped at 65 because none of the seven evidence dimensions reached CONFIRMED quality — there is no high-quality anchor for the analysis.
▸ SRC-CAP-004
Confidence is capped at 65 because telemetry is a single point-in-time snapshot with no ongoing monitoring — right-sizing needs usage patterns over time.
▸ RULE T2 — Right-Sizing Hard Cap
Right-sizing is limited to a moderate 25% reduction because telemetry is a single snapshot. More aggressive right-sizing needs 30+ days of monitoring data.
Decision Path
The step-by-step trace of how the recommendation was reached, in the system's own terms.
PRE-CHECK: hardware_dependency=false AND life_safety_classification=false → pre-check does not fire. RULE 1 (REPURCHASE): saas_alternative_available=true (notes + procurement confirm Infinitt Cloud PACS) AND saas_confidence_inferred=70 >= 50 AND hardware_dependency=false → RULE 1 FIRES → strategic_recommendation=REPURCHASE. CONFIDENCE SCORING: base=50; telemetry SINGLE (+6.00); dependency_mapping INFERRED (+2.50); procurement_contracts MISSING (+0.00); architecture_diagrams INFERRED (+1.75); security_compliance INFERRED (+1.25); business_criticality SINGLE (+2.00); dr_requirements MISSING (+0.00); pre-cap total=63.50. CAP EVALUATION: SRC-CAP-001 fires (phi_handled=true, baa_confirmed=UNKNOWN) → cap=40; SRC-CAP-002 fires (5/7 dimensions INFERRED/MISSING) → cap=55; SRC-CAP-003 fires (zero CONFIRMED) → cap=65; SRC-CAP-004 fires (snapshot_count=1, apm=false) → cap=65; lowest cap wins → SRC-CAP-001 → final_confidence=40. GATE (v4.1): final_confidence=40 < 60 AND strategic_recommendation=REPURCHASE (not RETAIN) → recommendation_status=PENDING_CONDITIONS; displayed_primary=REPURCHASE (gate does NOT force RETAIN under v4.1). CONDITIONS TO UNLOCK: (1) Confirm BAA with Infinitt for Cloud PACS; (2) Obtain and validate contract terms and data migration SLA; (3) Conduct formal dependency mapping for HL7 feed, DICOM modality integrations, and downstream viewer/dose consumers; (4) Collect APM telemetry to confirm utilization profile before right-sizing.
Cap Evaluation (all fired caps): SRC-CAP-001: PHI handled + BAA not confirmed → ceiling 40 (ceiling=40) ← BINDING SRC-CAP-002: ≥4 of 7 dimensions INFERRED/MISSING → ceiling 55 (ceiling=55) SRC-CAP-003: Zero CONFIRMED dimensions → ceiling 65 (ceiling=65) SRC-CAP-004: Single snapshot, no APM → ceiling 65 (ceiling=65)
Confidence Scoring Breakdown
How the confidence percentage was calculated, dimension by dimension.
Base score: 50 Evidence contributions: Telemetry SINGLE +6.00 Dependency Mapping INFERRED +2.50 Procurement Contracts MISSING +0.00 Architecture Diagrams INFERRED +1.75 Security Compliance INFERRED +1.25 Business Criticality SINGLE +2.00 Dr Requirements MISSING +0.00 Pre-cap score: 63.5 Binding cap applied: SRC-CAP-001 Final confidence: 40%
Intelligence Gaps
Specific things we don't yet know. Each gap is a reason the confidence score is held back.
BAA Status with Infinitt Cloud PACS
phi_handled=true and baa_confirmed=UNKNOWN triggers SRC-CAP-001, capping confidence at 40. No HIPAA-compliant SaaS transition can be executed without a signed BAA. This is the binding blocker for REPURCHASE execution.
Contract Intelligence — Infinitt Cloud PACS Terms
contract_intelligence=INSUFFICIENT_EVIDENCE means pricing, data migration terms, SLA, and exit provisions are unknown. repurchase_viable=INSUFFICIENT_EVIDENCE cannot be resolved without contract review.
Formal Dependency Mapping
All upstream and downstream dependencies (HL7 feed source, DICOM modalities, dose tracking consumers, network gateway routing) are inferred from VM naming patterns. No confirmed dependency map exists. Migration sequencing and cutover risk cannot be accurately assessed without this data.
APM Telemetry — Application-Layer Performance
apm_telemetry_available=false means application-layer latency, DICOM throughput, and concurrent user load are unknown. Right-sizing recommendation (25% reduction) and cloud tier selection carry elevated uncertainty without this data.
DR / RTO-RPO Requirements
No DR documentation cited. For a High-criticality clinical imaging system handling PHI, RTO/RPO requirements must be confirmed before cloud architecture can be finalized.
Latency Sensitivity Classification
latency_sensitive=UNKNOWN. PACS systems serving real-time diagnostic imaging may have latency constraints that affect SaaS viability or require dedicated connectivity (e.g., AWS Direct Connect, Azure ExpressRoute). Confirmation required.
DICOM Modality Integration Confirmation
Dental imaging acquisition devices (X-ray, CBCT) are inferred via industry pattern, not confirmed. If any modality has a hardware-level DICOM gateway dependency on the on-premises PACS, migration complexity increases significantly.
Technical Detail
A fuller technical explanation of the recommendation and its constraints.
Dependency agent confirms hardware_dependency=false and life_safety_classification=false, clearing the structural retain pre-check. Procurement agent identifies Infinitt Cloud PACS as a vendor-hosted SaaS alternative with saas_confidence_inferred=70, satisfying Rule 1's threshold of 50. The application notes explicitly reference 'Infinitt Cloud PACS — vendor-hosted option available,' corroborating the procurement finding. Rule 1 fires as the first match, yielding REPURCHASE as the strategic recommendation; however, SRC-CAP-001 applies because phi_handled=true and baa_confirmed=UNKNOWN, capping confidence at 40 and triggering PENDING_CONDITIONS status.
Secondary Path
The fallback option if the primary recommendation cannot proceed.
Secondary Path — REHOST
REHOST is the secondary disposition: all 12 VMs are powered-on Windows Server 2019 workloads with no hardware dependency, making a standard IaaS lift-and-shift viable if the SaaS path is blocked by BAA or contract constraints. Telemetry agent flags a 25% right-sizing opportunity (SRC-RS-002) applicable under either path.
Right-Sizing Notes
Why the cloud sizing estimate is set where it is, based on the telemetry available.
Fleet avg_mem_active_pct = 11.4% (< 25%), which under Rule T3 would indicate OVERSIZED (SRC-RS-003, 0.40 reduction). However, Rule T2 (single-snapshot hard rule) is unconditionally in effect: snapshot_count=1 AND apm_telemetry_available=false. SRC-RS-003 and SRC-RS-004 are FORBIDDEN. Tier is capped at SRC-RS-002 (MODERATE, 0.25 reduction). A single point-in-time snapshot cannot distinguish genuinely idle VMs from those sampled between burst cycles — this is the precise scenario T2 guards against. CPU utilization data is unavailable (cpu_avg_pct absent from all 12 VM records; fleet_avg_cpu_pct = null, DEF-RS-CPU-NULL applied). All 12 VMs run Windows Server 2019 (64-bit), EOL October 2026 — no EOL violations. No memory pressure detected across fleet. Application is phi_handling=true with baa_valid=null (BAA confirmation status UNKNOWN); synthesizer must evaluate BAA gap under SRC-CAP-001 governance. Confidence ceiling of 40 applied per SRC-CAP-001.
This tab shows the raw output of each analysis agent — the specialized components that examine telemetry, dependencies, contracts, and cost. It is the most detailed and most technical view. Most readers will find the Overview and Backlog tabs sufficient.
Telemetry Agent
Utilization Profilemoderate
Snapshot Count1
APM AvailableFalse
Hard Rule TriggeredTrue
Right-Sizing Opp.True
Agent Confidence40%
Telemetry — VM Counts
Powered Off VMs0
Flagged VMs0
Mem Active % (avg)11.4%
Data WindowNone
Inventory Source
Dependency Agent
Epic IntegrationFalse
PHI HandledTrue
Hardware DependencyFalse
Latency SensitiveUNKNOWN
Blast Radiushigh
Migration RiskMEDIUM
Conf (Inferred)72%
Conf (Confirmed)55%
Verificationsingle_source
Cloud Connectivity
ComplianceHIPAA / HITRUST
Upstream (2)
HL7 Feed Source (EHR / Scheduling / RIS): inferred: VM CDMINFHL701 and CDMINFHL7T01 are named with 'HL7' suffix, indicating an HL7 interface engine receiving…
Dental Imaging Acquisition Devices (X-ray / CBCT sensors): inferred: PACS systems structurally require upstream DICOM-sending modalities (dental X-ray units, panoramic sensors,…
Downstream (4)
Dental Clinic Workstations / Diagnostic Viewers: inferred: VM CDMINFULITE01 and CDMINFULITET01 named 'ULITE' consistent with Infinitt U-LITE thin-client viewer…
Dose Tracking / Radiation Management System: inferred: VMs CDMINFDOSE01 and CDMINFDOSET01 named 'DOSE' indicate a dose management module with potential downstream…
Network Gateway / External Routing (NETG): inferred: VMs CDMINFNETG01 and CDMINFNETGT01 named 'NETG' suggest a network gateway component routing DICOM or HL7…
IIS Web Front-End Consumers: inferred: VMs CDMINFIIS01 and CDMINFIIST01 named 'IIS' indicate a web application tier serving downstream browser-based…
Procurement Agent
SaaS Available—
SaaS VendorInfinitt
SaaS ProductInfinitt Cloud PACS
Repurchase ViableINSUFFICIENT_EVIDENCE
BAA ConfirmedUNKNOWN
SaaS Confidence
Inferred Score70%
Confirmed Score0%
Verificationsingle_source
AWS Equiv.None
Provisioning Agent (Python engine — not LLM)
Current VMs12
Recommended VMs9
Right-Sizing25% reduction
Current TCO/yr$78,000
Projected Cloud TCO$46,800
Migration One-Time$9,360
Dual-Running Cost$41,600
-$19,760
Year-1 Net
$31,200
Steady-State
$42,640
3-Year
AWS TargetAmazon EC2
Azure TargetAzure Virtual Machines
Cost Modeldefault_multipliers
DEF: {'default_id': 'DEF-WL-001', 'field': 'workload_type', 'value': 'app_server', 'reason': 'workload_type indeterminable'}
DEF: {'default_id': 'DEF-MIG-001', 'field': 'migration_factor', 'value': 0.2, 'reason': 'migration_complexity not provided'}
DEF: {'default_id': 'DEF-COST-001', 'field': 'cost_model_basis', 'value': 'default_multipliers', 'reason': 'no customer cloud quote provided; using SRC-TCO-* and SRC-MULT-* registry defaults'}
ATNA Audit Repository
Vendor: Sectra · 6 VMs · Criticality: High · HIPAA / HITRUST — PHI audit log reposito
REHOST
40%
Confidence
28%
Data Quality
ROI: Year-1 Net: -$5,893 · Steady-State: $18,200/yr · 3-Year: $30,507
Overview
Backlog
Provenance
Technical Details
Agents
Recommendation
REHOST ⏳ PENDING CONDITIONS
SRC-CAP-001 Binding cap · Ceiling: 40% · Threshold: 60%
Secondary: RETAIN
Confidence
40%
Viable, but specific gaps must close first.
ROI Summary (Python-computed, not LLM)
-$5,893
Year-1 Net
$18,200
Steady-State /yr
$30,507
3-Year Total
ROI variance: ±60%
⚙ Financials computed by provisioning_engine_v3 (Python) — not LLM-generated
In plain terms: migrating costs money up front — Year 1 is a net outflow of $5,893. From Year 2 onward the app saves $18,200 every year. The 3-Year total nets these against each other.
Finding
The ATNA Audit Repository is a Sectra-managed, VM-based compliance workload with no hardware dependency, making it a viable candidate for cloud rehosting. However, the BAA status for this PHI-handling system is unconfirmed, and the latency sensitivity of its upstream clinical imaging and scheduling integrations has not been validated — both must be resolved before migration can proceed. Once those conditions are met, a standard IaaS lift-and-shift with right-sizing (25% reduction) is projected to save approximately $18,200 annually.
In plain terms: see the recommendation and conditions above.
Conditions to Unlock (3 items)
This recommendation becomes executable once these conditions are confirmed:
☐ Verify Business Associate Agreement with Sectra Covers Cloud Hosting
Owner: Compliance Officer / Legal Counsel · Timeline: 14 days · Unlocks: compliance_close, cap_resolution, migration_commit
☐ Confirm Whether Real-Time Audit Logging Has Strict Timing Requirements
Owner: Radiology Application Owner (Wendy Ketchum) / Network Architect · Timeline: 10 days · Unlocks: migration_commit, operational_safety
☐ Install Application Performance Monitoring to Validate Server Sizing
Owner: Infrastructure / Platform Engineering · Timeline: 21 days · Unlocks: cost_clarity, migration_commit
See Backlog tab for full action backlog and P2/P3 items.
This is the list of work needed to move this recommendation forward. The 3 P1 items are the priority — they unlock the recommendation. P2 and P3 items improve confidence and reduce risk but are not blockers. 'Story points' measure effort, not time or cost — see the Methodology tab for how they work.
40%
Current
60%
Threshold
55%
Ceiling
34
Story Points
Why This Score
Confidence is fixed at 40 by the binding SRC-CAP-001 cap (phi_handled=true, baa_confirmed=UNKNOWN); the pre-cap score of 64.5 reflects a viable REHOST candidate, but no amount of secondary evidence collection can lift the score until the BAA is confirmed with Sectra. Four concurrent caps (SRC-CAP-001 through SRC-CAP-004) are active, with SRC-CAP-001 as the sole binding constraint at this stage; resolving it is the necessary and sufficient first step to advancing confidence toward the 60-point execution threshold.
Proceeding with cloud migration planning without a confirmed Sectra BAA exposes MUSC to HIPAA enforcement risk under 45 CFR 164.308(b) for a PHI-handling audit repository with no documented business associate relationship for the cloud deployment model.
Action Backlog — 6 Items · 34 Story Points · Ceiling: 55%
APP-021-001 Confirm BAA with Sectra for PHI-Handling Cloud Deployment
P1 BAA/Legal
Obtain and validate a signed Business Associate Agreement (BAA) with Sectra that explicitly covers cloud-hosted deployment of the ATNA Audit Repository. Confirm the BAA scope includes the MUSC legal entity, the cloud target environment (AWS/Azure/GCP/OCI), and all three workloads sharing the SAHUBDB cluster (ATNA, SCH, WFA). This is the single binding blocker: SRC-CAP-001 caps confidence at 40 unt
+15% confidence -0% ROI variance 👤 📋 · 8 pts
APP-021-002 Validate Latency Sensitivity of ATNA ARR Upstream Integrations
P1 Dependency Mapping
Engage Radiology team owner (Wendy Ketchum) and Sectra to determine whether the ATNA Audit Record Repository enforces synchronous audit-commit-before-clinical-action semantics or operates in fully asynchronous syslog/TLS push mode per the IHE ATNA profile baseline. Enumerate all upstream IHE ATNA-participating nodes (modality worklist, image archive, viewer nodes, SCH/WFA) sending audit events to
+0% confidence -0% ROI variance 👤 📋 · 5 pts
APP-021-003 Deploy 30-Day APM Telemetry Collection for Right-Sizing Validation
P1 Telemetry
Deploy application-layer performance monitoring (APM) across all 6 ATNA Repository VMs (SAATNADB01/02, SACONATNA01/02, SAHUBDB01/02) to capture a minimum 30-day time-series of CPU utilization, memory active, audit event ingestion rates, DB transaction volumes, and AoAG replication lag. Current right-sizing is capped at MODERATE (SRC-RS-002, 25% reduction) due to single-snapshot RVTools data; APM t
+0% confidence -0% ROI variance 👤 📋 · 8 pts
APP-021-004 Obtain Sectra Contract and Confirm Cloud Licensing Terms
P2 Procurement
Retrieve the current Sectra software license agreement and support contract for the ATNA Audit Repository (and co-licensed SCH/WFA components sharing SAHUBDB). Confirm whether the existing agreement permits cloud-hosted deployment or requires a separate cloud SKU, addendum, or re-licensing. Document ACV, renewal date, and any cloud-restriction clauses. Sectra is a Swedish vendor; confirm the BAA a
+0% confidence -0% ROI variance 👤 📋 · 5 pts
APP-021-005 Document IHE ATNA Network Topology and Migration Wave Coupling
P2 Architecture
Produce a confirmed architecture diagram of the IHE ATNA network topology: enumerate all audit-source nodes (modality worklist, image archive, viewer nodes, SCH, WFA) sending syslog/TLS audit events to SACONATNA01/02; document firewall rules, network paths, and port requirements. Critically, confirm whether ATNA, SCH, and WFA applications sharing SAHUBDB01/02 can be decoupled for independent migra
+0% confidence -0% ROI variance 👤 📋 · 5 pts
APP-021-006 Document Formal RTO/RPO Requirements for ATNA Audit Repository
P3 Architecture
Obtain formal, documented RTO and RPO requirements for the ATNA Audit Repository from the MUSC Radiology team and compliance/risk office. Current DR posture is inferred from dual-DC VM placement (MUSC-ART + MUSC-USB AoAG topology); no formal SLA document exists. For HIPAA audit control continuity (45 CFR 164.312(b)), the cloud AZ mapping must preserve equivalent or superior DR posture. Confirm whe
+0% confidence -0% ROI variance 👤 📋 · 3 pts
This tab shows the evidence behind the recommendation. The seven dimensions are scored by quality — CONFIRMED evidence counts fully, INFERRED evidence counts less. The validator table confirms each analysis step passed its checks. Green means the step is sound.
Evidence Classification — 7 Dimensions
Dimension Classification Multiplier Contribution
Telemetry SINGLE 0.50 6.00
Dependency Mapping INFERRED 0.25 2.50
Procurement Contracts MISSING 0.00 0.00
Architecture Diagrams INFERRED 0.25 1.75
Security Compliance INFERRED 0.25 1.25
Business Criticality SINGLE 0.50 2.00
Dr Requirements INFERRED 0.25 1.00
Applied Cap: SRC-CAP-001
Validation Reports — Per Agent
Agent Status Provenance Counts Defaults Used Rule Failures
telemetry PASS
SRC:2CUST:8COMP:5DEF:2
 None ✓ None
dependency PASS
SRC:12COMP:5DEF:3
 None ✓ None
procurement PASS
None ✓ None
provisioning PASS_WITH_DEFAULTS
SRC:7CUST:1COMP:9DEF:1
• {'default_id': 'DEF-WL-001', 'field': 'workload_type', 'value': 'app_server', 'r
• {'default_id': 'DEF-MIG-001', 'field': 'migration_factor', 'value': 0.2, 'reason
 ✓ None
synthesizer PASS
None ✓ None
confidence_advisor PASS
COMP:15
 None ✓ None
This section contains the full technical reasoning behind the recommendation: the rules that fired, the decision path, the confidence math, and the open intelligence gaps. It is intended for analysts and engineers. Executive readers can rely on the Overview tab.
Rules Applied
These are the specific governance rules that produced this recommendation.
▸ RULE 6 — REHOST
No other qualifying path applies — the default for virtual workloads is to rehost to cloud IaaS.
▸ SRC-CAP-001 (binding)
Confidence is capped at 40 because the app handles PHI and the vendor's Business Associate Agreement is not confirmed. PHI cannot move to a cloud vendor without a confirmed BAA.
▸ SRC-CAP-002
Confidence is capped at 55 because more than half of the seven evidence dimensions are inferred or missing — the conclusion is data-thin.
▸ SRC-CAP-003
Confidence is capped at 65 because none of the seven evidence dimensions reached CONFIRMED quality — there is no high-quality anchor for the analysis.
▸ SRC-CAP-004
Confidence is capped at 65 because telemetry is a single point-in-time snapshot with no ongoing monitoring — right-sizing needs usage patterns over time.
▸ RULE T2 — Right-Sizing Hard Cap
Right-sizing is limited to a moderate 25% reduction because telemetry is a single snapshot. More aggressive right-sizing needs 30+ days of monitoring data.
Decision Path
The step-by-step trace of how the recommendation was reached, in the system's own terms.
PRE-CHECK: hardware_dependency=false, life_safety_classification=false — structural blocker does not fire. RULE 1 (REPURCHASE): saas_available=false, saas_confidence_inferred=10 — does not fire. RULE 2 (RETIRE): no decommission annotations, no named successor — does not fire. RULE 3 (REFACTOR): vendor COTS (Sectra), not internally developed — does not fire. RULE 4 (REHOST): VM-based=true, hardware_dependency=false=true, latency_sensitive=UNKNOWN — UNKNOWN does not confirm a constraint; per v4.1 REFUSAL guidance, UNKNOWN produces PENDING_CONDITIONS not RETAIN — RULE 4 FIRES → strategic_recommendation=REHOST. CONFIDENCE GATE: pre-cap=64.5; SRC-CAP-001 (phi_handled=true, baa_confirmed=UNKNOWN) caps at 40; SRC-CAP-002 (5/7 dimensions INFERRED/MISSING) caps at 55; SRC-CAP-003 (zero CONFIRMED) caps at 65; SRC-CAP-004 (snapshot_count=1, apm=false) caps at 65; lowest binding cap = SRC-CAP-001 → final_confidence=40. Gate: 40 < 60 AND strategic=REHOST → recommendation_status=PENDING_CONDITIONS, displayed_primary=REHOST (gate does NOT force RETAIN in v4.1).
Cap Evaluation (all fired caps): SRC-CAP-001: PHI handled + BAA not confirmed → ceiling 40 (ceiling=40) ← BINDING SRC-CAP-002: ≥4 of 7 dimensions INFERRED/MISSING → ceiling 55 (ceiling=55) SRC-CAP-003: Zero CONFIRMED dimensions → ceiling 65 (ceiling=65) SRC-CAP-004: Single snapshot, no APM → ceiling 65 (ceiling=65)
Confidence Scoring Breakdown
How the confidence percentage was calculated, dimension by dimension.
Base score: 50 Evidence contributions: Telemetry SINGLE +6.00 Dependency Mapping INFERRED +2.50 Procurement Contracts MISSING +0.00 Architecture Diagrams INFERRED +1.75 Security Compliance INFERRED +1.25 Business Criticality SINGLE +2.00 Dr Requirements INFERRED +1.00 Pre-cap score: 64.5 Binding cap applied: SRC-CAP-001 Final confidence: 40%
Intelligence Gaps
Specific things we don't yet know. Each gap is a reason the confidence score is held back.
BAA Confirmation — Sectra ATNA Repository
phi_handled=true with baa_confirmed=UNKNOWN triggers SRC-CAP-001, capping confidence at 40 and blocking EXECUTABLE status. BAA must be confirmed with Sectra before any cloud migration of this PHI audit repository can proceed. This is the single binding blocker.
Latency Sensitivity — Sectra PACS / SCH / WFA Upstream Integration
latency_sensitive=UNKNOWN for a system receiving real-time IHE ATNA syslog/TLS audit events from clinical imaging modalities and scheduling systems. If latency constraints are confirmed, cloud network path to on-premises Sectra PACS must be validated (Direct Connect / ExpressRoute) before migration.
APM Telemetry Absent — Right-Sizing Risk
apm_telemetry_available=false with snapshot_count=1 means right-sizing recommendation (SRC-RS-002, 25% reduction, 6→4 VMs) is based on a single RVTools memory snapshot at 10.8% avg active. Application-layer throughput, audit event ingestion rates, and DB transaction volumes are unknown. Over-aggressive right-sizing could impact audit log retention SLAs.
Procurement Contract Intelligence — Sectra Licensing
contract_intelligence=INSUFFICIENT_EVIDENCE. Sectra licensing terms for cloud deployment are unknown. Some Sectra agreements restrict cloud hosting or require separate cloud SKUs. Contract review required before migration commitment.
Architecture Diagrams — IHE ATNA Network Topology
No architecture diagrams available. The full set of IHE ATNA-compliant audit sources (modality nodes, image archive, viewer nodes) sending to this repository is inferred from industry pattern, not confirmed. Blast radius estimate is HIGH — migration without confirmed topology risks audit log gaps for HIPAA compliance.
DR Requirements — Dual-DC Topology Validation
DR requirements inferred from dual-DC VM placement (MUSC-ART + MUSC-USB). Formal RTO/RPO requirements for the ATNA audit repository are not documented. Cloud AZ mapping must preserve equivalent DR posture for HIPAA audit control continuity.
Technical Detail
A fuller technical explanation of the recommendation and its constraints.
Telemetry confirms 6 powered-on VMs across dual datacenters with moderate utilization (10.8% avg memory active) and no hardware dependency, satisfying Rule 4 REHOST criteria. Procurement analysis finds no SaaS alternative (saas_available=false, saas_confidence_inferred=10), eliminating Rule 1. No decommission annotations or named successors eliminate Rule 2. Sectra COTS classification eliminates Rule 3. The binding constraint is SRC-CAP-001: phi_handled=true with baa_confirmed=UNKNOWN caps confidence at 40, placing this recommendation in PENDING_CONDITIONS status until BAA is confirmed and latency sensitivity of the Sectra PACS/SCH/WFA upstream cluster is validated.
Secondary Path
The fallback option if the primary recommendation cannot proceed.
Secondary Path — RETAIN
Secondary disposition RETAIN reflects the operational reality that if BAA cannot be confirmed or latency constraints prove prohibitive for the IHE ATNA audit stream, the workload must remain on-premises within the existing dual-DC topology.
Right-Sizing Notes
Why the cloud sizing estimate is set where it is, based on the telemetry available.
Fleet avg_mem_active_pct of 10.8% would classify as OVERSIZED (SRC-RS-003) under RULE T3 (threshold: <25%), but RULE T2 hard-caps the maximum tier at MODERATE (SRC-RS-002) because snapshot_count=1 and apm_telemetry_available=false. Single point-in-time data cannot distinguish genuinely idle VMs from those sampled between burst cycles. OVERSIZED (SRC-RS-003) and ZOMBIE (SRC-RS-004) tiers are FORBIDDEN for this run. Additionally, VM notes confirm an Always-On Availability Group (AoAG) topology with active/passive node pairs split across MUSC-USB (primary) and MUSC-ART (secondary) datacenters. Passive nodes SAATNADB02 (5.0%), SACONATNA02 (4.0%), and SAHUBDB02 (10.0%) are architecturally expected to show suppressed memory utilization at rest and must not be sized independently of their active counterparts. This topology pattern further reinforces the inadequacy of single-snapshot assessment. A minimum of 30-day APM telemetry or multi-snapshot RVTools series is required before any right-sizing action is taken on this High-criticality, PHI-handling application.
This tab shows the raw output of each analysis agent — the specialized components that examine telemetry, dependencies, contracts, and cost. It is the most detailed and most technical view. Most readers will find the Overview and Backlog tabs sufficient.
Telemetry Agent
Utilization Profilemoderate
Snapshot Count1
APM AvailableFalse
Hard Rule TriggeredTrue
Right-Sizing Opp.True
Agent Confidence44%
Telemetry — VM Counts
Powered Off VMs0
Flagged VMs0
Mem Active % (avg)10.8%
Data WindowNone
Inventory Source
Dependency Agent
Epic IntegrationFalse
PHI HandledTrue
Hardware DependencyFalse
Latency SensitiveUNKNOWN
Blast Radiushigh
Migration RiskMEDIUM
Conf (Inferred)72%
Conf (Confirmed)58%
Verificationsingle_source
Cloud Connectivity
ComplianceHIPAA / HITRUST
Upstream (3)
Sectra PACS / RIS (Radiology): inferred: rvtools.vm_notes SAATNADB01/02 'Team: Radiology'; ATNA audit repository receives audit events from clinical…
SCH / WFA System (Scheduling / Workflow Automation): confirmed: rvtools.vm_notes SAHUBDB01/02 explicitly state 'SCH/ATNA/WFA DB AoAG Node 1/2' — shared DB cluster indicates…
IHE ATNA-compliant audit sources (network nodes): inferred: IHE ATNA profile requires all participating nodes to send syslog/TLS audit messages to the repository; in a…
Downstream (3)
Compliance / Audit Reporting Systems: inferred: ATNA repositories are consumed by compliance officers and audit tools for HIPAA access log review; standard…
Epic EHR (potential audit consumer): inferred: catalog.epic_integration=false confirms no direct Epic integration feed; however Epic environments at covered…
Security / SIEM Platform: inferred: ATNA audit logs are a standard SIEM feed source in healthcare environments for anomaly detection and breach…
Procurement Agent
SaaS Available—
SaaS Vendor
SaaS Product
Repurchase ViableFalse
BAA ConfirmedUNKNOWN
SaaS Confidence
Inferred Score10%
Confirmed Score0%
Verificationinferred_only
AWS Equiv.None
Provisioning Agent (Python engine — not LLM)
Current VMs6
Recommended VMs4
Right-Sizing25% reduction
Current TCO/yr$39,000
Projected Cloud TCO$20,800
Migration One-Time$4,160
Dual-Running Cost$19,933
-$5,893
Year-1 Net
$18,200
Steady-State
$30,507
3-Year
AWS TargetAmazon EC2
Azure TargetAzure Virtual Machines
Cost Modeldefault_multipliers
DEF: {'default_id': 'DEF-WL-001', 'field': 'workload_type', 'value': 'app_server', 'reason': 'workload_type indeterminable'}
DEF: {'default_id': 'DEF-MIG-001', 'field': 'migration_factor', 'value': 0.2, 'reason': 'migration_complexity not provided'}
DEF: {'default_id': 'DEF-COST-001', 'field': 'cost_model_basis', 'value': 'default_multipliers', 'reason': 'no customer cloud quote provided; using SRC-TCO-* and SRC-MULT-* registry defaults'}
Butterfly Link
Vendor: Butterfly Network · 3 VMs · Criticality: High · HIPAA / HITRUST — phi_handled=true; BAA
REPURCHASE
40%
Confidence
38%
Data Quality
ROI: Year-1 Net: -$2,947 · Steady-State: $9,100/yr · 3-Year: $15,253
Overview
Backlog
Provenance
Technical Details
Agents
Recommendation
REPURCHASE ⏳ PENDING CONDITIONS
SRC-CAP-001 Binding cap · Ceiling: 40% · Threshold: 60%
Secondary: REHOST
Confidence
40%
Viable, but specific gaps must close first.
ROI Summary (Python-computed, not LLM)
-$2,947
Year-1 Net
$9,100
Steady-State /yr
$15,253
3-Year Total
ROI variance: ±50%
⚙ Financials computed by provisioning_engine_v3 (Python) — not LLM-generated
In plain terms: migrating costs money up front — Year 1 is a net outflow of $2,947. From Year 2 onward the app saves $9,100 every year. The 3-Year total nets these against each other.
Finding
Butterfly Link is a cloud-connected gateway that already routes ultrasound data to Butterfly Network's own cloud platform — the vendor offers a fully hosted SaaS alternative (Butterfly Cloud) that would eliminate all three on-premises gateway VMs. However, before committing to the SaaS transition, MUSC must execute a Business Associate Agreement with Butterfly Network and obtain formal contract terms, because this workload handles protected health information and no BAA is currently on file. Once those conditions are met, this is a straightforward vendor-managed migration with meaningful infrastructure savings.
In plain terms: this app should move to the vendor's cloud product, but a few specific items must be confirmed first. See Conditions to Unlock below.
Conditions to Unlock (1 item)
This recommendation becomes executable once these conditions are confirmed:
☐ Obtain Business Associate Agreement with Butterfly Network
Owner: Compliance / Legal · Timeline: 21 days · Unlocks: compliance_close, cap_resolution, migration_commit
See Backlog tab for full action backlog and P2/P3 items.
This is the list of work needed to move this recommendation forward. The 1 P1 item is the priority — it unlocks the recommendation. P2 and P3 items improve confidence and reduce risk but are not blockers. 'Story points' measure effort, not time or cost — see the Methodology tab for how they work.
40%
Current
60%
Threshold
55%
Ceiling
27
Story Points
Why This Score
Confidence is capped at 40 by SRC-CAP-001 (phi_handled=true, baa_confirmed=UNKNOWN), which is the single lowest and most binding of four triggered caps (SRC-CAP-001 at 40, SRC-CAP-002 at 55, SRC-CAP-003 at 65, SRC-CAP-004 at 65); the pre-cap score of 66.0 reflects a well-understood application architecture, but zero confirmed dimensions and missing procurement and DR data prevent any score above 40 until the BAA is executed. The REPURCHASE recommendation is structurally sound — the vendor SaaS path is confirmed by two independent catalog fields and all three VM notes — but it cannot advance to EXECUTABLE status until the BAA compliance gap is resolved and contract intelligence is obtained.
Proceeding with any migration action before BAA execution exposes MUSC to HIPAA enforcement risk for unauthorized PHI disclosure to a business associate without a compliant agreement, and the absence of contract intelligence means the ROI projection carries a ±60% variance that could invert the business case.
Action Backlog — 6 Items · 27 Story Points · Ceiling: 55%
APP-022-001 Execute BAA with Butterfly Network to resolve SRC-CAP-001 and unlock REPURCHASE disposition
P1 BAA/Legal
Butterfly Link handles PHI (phi_handled=true) and routes HL7 messages and ultrasound image data to Butterfly Network's vendor-hosted cloud. No BAA with Butterfly Network is on file (baa_valid=null → baa_confirmed=UNKNOWN). Under HIPAA, a BAA is a mandatory prerequisite before any PHI may be transmitted to or processed by a business associate. SRC-CAP-001 caps confidence at 40 until baa_confirmed=t
+15% confidence -0% ROI variance 👤 📋 · 8 pts
APP-022-002 Obtain Butterfly Cloud contract terms, SaaS pricing, and data migration scope to resolve procurement intelligence gap
P2 Procurement
procurement_contracts=MISSING and saas_confidence_confirmed=0. The REPURCHASE recommendation to Butterfly Cloud cannot be financially validated without contract terms. Required artifacts: (1) Butterfly Cloud SaaS subscription pricing for MUSC's patient volume and probe fleet, (2) data migration scope and any one-time transition fees, (3) contract renewal terms and SLA commitments (uptime, support
+0% confidence -0% ROI variance 👤 📋 · 5 pts
APP-022-003 Identify and document upstream HL7 source system feeding butterflyhl7 VM
P2 Dependency Mapping
The VM named 'butterflyhl7' implies an active HL7 interface engine receiving orders, ADT events, or study metadata from an upstream EHR or order entry system. This upstream source system is not identified in the application catalog or RVTools data. Without this mapping, the migration cutover plan cannot sequence the HL7 feed handoff correctly — a gap that could cause clinical workflow disruption (
+0% confidence -0% ROI variance 👤 📋 · 3 pts
APP-022-004 Confirm PACS/RIS downstream dependency and validate Butterfly Cloud routing continuity
P2 Dependency Mapping
Ultrasound image gateways in healthcare environments typically forward DICOM studies to a PACS (Picture Archiving and Communication System) or RIS (Radiology Information System). This downstream path is currently inferred from industry context only — it is not confirmed in the application catalog or RVTools notes. If Butterfly Cloud does not absorb PACS/DICOM routing natively, a residual on-premis
+0% confidence -0% ROI variance 👤 📋 · 3 pts
APP-022-005 Deploy 30-day APM telemetry collection to validate utilization profile and resolve SRC-CAP-004
P2 Telemetry
apm_telemetry_available=false and snapshot_count=1 trigger SRC-CAP-004 (confidence ceiling 65) and the Rule T2 hard rule, which prohibits ZOMBIE and OVERSIZED right-sizing classifications. The current fleet avg_mem_active_pct of 4.0% falls below the 5% ZOMBIE threshold, but a single point-in-time snapshot cannot distinguish a genuinely idle VM from one sampled between burst cycles. Deploy lightwei
+0% confidence -0% ROI variance 👤 📋 · 5 pts
APP-022-006 Document DR/RTO requirements and validate Butterfly Cloud SLA against MUSC recovery objectives
P3 Architecture
dr_requirements=MISSING for a High-criticality, PHI-handling workload. No RTO or RPO targets are documented for Butterfly Link, and the Butterfly Cloud SaaS SLA has not been validated against MUSC's recovery objectives. Before committing to the SaaS transition, MUSC must: (1) obtain RTO/RPO requirements from clinical operations and IT leadership for the ultrasound imaging workflow, (2) request But
+0% confidence -0% ROI variance 👤 📋 · 3 pts
This tab shows the evidence behind the recommendation. The seven dimensions are scored by quality — CONFIRMED evidence counts fully, INFERRED evidence counts less. The validator table confirms each analysis step passed its checks. Green means the step is sound.
Evidence Classification — 7 Dimensions
Dimension Classification Multiplier Contribution
Telemetry SINGLE 0.50 6.00
Dependency Mapping SINGLE 0.50 5.00
Procurement Contracts MISSING 0.00 0.00
Architecture Diagrams INFERRED 0.25 1.75
Security Compliance INFERRED 0.25 1.25
Business Criticality SINGLE 0.50 2.00
Dr Requirements MISSING 0.00 0.00
Applied Cap: SRC-CAP-001
Validation Reports — Per Agent
Agent Status Provenance Counts Defaults Used Rule Failures
telemetry PASS
SRC:2CUST:9COMP:4DEF:2
 None ✓ None
dependency PASS
SRC:8CUST:1COMP:4DEF:3
 None ✓ None
procurement PASS
None ✓ None
provisioning PASS_WITH_DEFAULTS
SRC:7CUST:1COMP:9DEF:1
• {'default_id': 'DEF-WL-001', 'field': 'workload_type', 'value': 'app_server', 'r
• {'default_id': 'DEF-MIG-001', 'field': 'migration_factor', 'value': 0.2, 'reason
 ✓ None
synthesizer PASS
None ✓ None
confidence_advisor PASS
COMP:34
 None ✓ None
This section contains the full technical reasoning behind the recommendation: the rules that fired, the decision path, the confidence math, and the open intelligence gaps. It is intended for analysts and engineers. Executive readers can rely on the Overview tab.
Rules Applied
These are the specific governance rules that produced this recommendation.
▸ RULE 1 — REPURCHASE
A vendor SaaS replacement is available and credible, and the app has no hardware constraint — so the recommendation is to replace it with the vendor's cloud service.
▸ SRC-CAP-001 (binding)
Confidence is capped at 40 because the app handles PHI and the vendor's Business Associate Agreement is not confirmed. PHI cannot move to a cloud vendor without a confirmed BAA.
▸ SRC-CAP-002
Confidence is capped at 55 because more than half of the seven evidence dimensions are inferred or missing — the conclusion is data-thin.
▸ SRC-CAP-003
Confidence is capped at 65 because none of the seven evidence dimensions reached CONFIRMED quality — there is no high-quality anchor for the analysis.
▸ SRC-CAP-004
Confidence is capped at 65 because telemetry is a single point-in-time snapshot with no ongoing monitoring — right-sizing needs usage patterns over time.
▸ RULE T2 — Right-Sizing Hard Cap
Right-sizing is limited to a moderate 25% reduction because telemetry is a single snapshot. More aggressive right-sizing needs 30+ days of monitoring data.
Decision Path
The step-by-step trace of how the recommendation was reached, in the system's own terms.
PRE-CHECK: hardware_dependency=false, life_safety_classification=false — structural blocker does not fire. RULE 1 (REPURCHASE): saas_alternative_available=true (CONFIRMED: catalog.notes + rvtools.vms[*].note), saas_confidence_inferred=72 >= 50, hardware_dependency=false — RULE 1 FIRES. strategic_recommendation=REPURCHASE. CONFIDENCE SCORING: base=50, telemetry(SINGLE)=+6.0, dependency_mapping(SINGLE)=+5.0, procurement_contracts(MISSING)=+0.0, architecture_diagrams(INFERRED)=+1.75, security_compliance(INFERRED)=+1.25, business_criticality(SINGLE)=+2.0, dr_requirements(MISSING)=+0.0 → pre_cap=66.0. CAP EVALUATION: SRC-CAP-001 triggered (phi_handled=true, baa_confirmed=UNKNOWN) → ceiling=40; SRC-CAP-002 triggered (4/7 dims INFERRED/MISSING) → ceiling=55; SRC-CAP-003 triggered (zero CONFIRMED) → ceiling=65; SRC-CAP-004 triggered (snapshot_count=1, apm=false) → ceiling=65. Lowest cap: SRC-CAP-001=40. final_confidence=40. GATE (v4.1): final_confidence=40 < 60 AND strategic_recommendation=REPURCHASE (not RETAIN) → recommendation_status=PENDING_CONDITIONS, displayed_primary=REPURCHASE (gate does NOT force RETAIN in v4.1), confidence_gate_applied=true.
Cap Evaluation (all fired caps): SRC-CAP-001: PHI handled + BAA not confirmed → ceiling 40 (ceiling=40) ← BINDING SRC-CAP-002: ≥4 of 7 dimensions INFERRED/MISSING → ceiling 55 (ceiling=55) SRC-CAP-003: Zero CONFIRMED dimensions → ceiling 65 (ceiling=65) SRC-CAP-004: Single snapshot, no APM → ceiling 65 (ceiling=65)
Confidence Scoring Breakdown
How the confidence percentage was calculated, dimension by dimension.
Base score: 50 Evidence contributions: Telemetry SINGLE +6.00 Dependency Mapping SINGLE +5.00 Procurement Contracts MISSING +0.00 Architecture Diagrams INFERRED +1.75 Security Compliance INFERRED +1.25 Business Criticality SINGLE +2.00 Dr Requirements MISSING +0.00 Pre-cap score: 66.0 Binding cap applied: SRC-CAP-001 Final confidence: 40%
Intelligence Gaps
Specific things we don't yet know. Each gap is a reason the confidence score is held back.
BAA with Butterfly Network — absent
phi_handled=true with baa_confirmed=UNKNOWN triggers SRC-CAP-001 (confidence ceiling 40). BAA execution is a legal prerequisite for any SaaS transition involving PHI. This is the binding unlock condition for EXECUTABLE status.
Contract intelligence — insufficient
procurement.contract_intelligence=INSUFFICIENT_EVIDENCE and saas_confidence_confirmed=0. SaaS transition terms, data migration scope, and pricing for Butterfly Cloud are unknown. Cannot validate ROI projection without contract data.
HL7 source system not identified
VM 'butterflyhl7' implies an HL7 interface engine but the upstream EHR or order entry system feeding it is not named in the catalog. Migration cutover plan requires mapping this dependency to avoid HL7 feed disruption.
PACS/RIS downstream dependency unconfirmed
Ultrasound image gateways typically forward studies to PACS or RIS; this downstream path is inferred from industry context only. If Butterfly Cloud does not absorb PACS routing, a residual on-prem integration component may be required post-migration.
APM telemetry absent — single RVTools snapshot
apm_telemetry_available=false and snapshot_count=1 limit right-sizing confidence (SRC-CAP-004, ceiling 65). Right-sizing tier MODERATE (SRC-RS-002, 25% reduction) is the maximum permissible under hard rule. Additional snapshots or APM data would validate the 4% avg memory utilization reading.
DR/RTO requirements — not documented
No DR requirements data available for a High-criticality PHI-handling workload. SaaS SLA from Butterfly Cloud must be validated against MUSC's RTO/RPO requirements before transition.
Latency sensitivity — UNKNOWN
latency_sensitive=UNKNOWN. If real-time ultrasound image streaming has latency constraints that the vendor SaaS cannot meet from MUSC's network topology, the REPURCHASE path may require network optimization or a hybrid gateway model.
Technical Detail
A fuller technical explanation of the recommendation and its constraints.
Rule 1 (REPURCHASE) fires: saas_alternative_available=true (Butterfly Cloud confirmed in catalog notes and all three RVTools VM notes), saas_confidence_inferred=72 >= 50, and hardware_dependency=false. The application is architecturally a transitional gateway — its sole confirmed downstream dependency is the vendor's own SaaS platform, making the on-premises VMs structurally redundant once the SaaS transition is executed. Confidence is capped at 40 by SRC-CAP-001 (phi_handled=true, baa_confirmed=UNKNOWN), triggering PENDING_CONDITIONS status; the strategic recommendation remains REPURCHASE with BAA execution and contract validation as unlock conditions. Right-sizing analysis (SRC-RS-002) confirms 25% over-provisioning across the 3-VM fleet at 4% average memory utilization.
Secondary Path
The fallback option if the primary recommendation cannot proceed.
Secondary Path — REHOST
REHOST is the secondary disposition if BAA and contract conditions cannot be resolved within the planning horizon — the VMs are current-generation Windows Server 2022 with no hardware dependency and would lift-and-shift cleanly to IaaS.
Right-Sizing Notes
Why the cloud sizing estimate is set where it is, based on the telemetry available.
Fleet avg_mem_active_pct=4.0% (butterflygtwy=3.0%, butterflyhl7=2.0%, butterflytst=7.0%). Raw utilization falls below the 5% ZOMBIE threshold (SRC-RS-004) under Rule T3. However, Rule T2 hard rule fires unconditionally: snapshot_count=1 AND apm_telemetry_available=false. SRC-RS-003 (OVERSIZED) and SRC-RS-004 (ZOMBIE) are FORBIDDEN. Tier is capped at SRC-RS-002 MODERATE (0.25 reduction). A single point-in-time snapshot cannot distinguish a genuinely idle VM from one sampled between burst cycles — this is precisely the unsafe reasoning pattern Rule T2 is designed to prevent. 30+ day APM telemetry is required before any escalation to ZOMBIE classification is permissible. Rule T1 did not fire: application_type='Cloud-Connected Gateway' is not 'Hardware-Integrated' and life_safety_classification is absent from catalog. No EOL OS violations detected: all three VMs run Windows Server 2022 (64-bit), which is within mainstream support. CPU utilization data is unavailable; cpu_avg_pct is null on all 3 of 3 VMs and no APM feed was provided. phi_handling=true is noted for downstream synthesizer; no EOL flags exist to escalate under Rule T4.
This tab shows the raw output of each analysis agent — the specialized components that examine telemetry, dependencies, contracts, and cost. It is the most detailed and most technical view. Most readers will find the Overview and Backlog tabs sufficient.
Telemetry Agent
Utilization Profilemoderate
Snapshot Count1
APM AvailableFalse
Hard Rule TriggeredTrue
Right-Sizing Opp.True
Agent Confidence45%
Telemetry — VM Counts
Powered Off VMs0
Flagged VMs0
Mem Active % (avg)4.0%
Data WindowNone
Inventory Source
Dependency Agent
Epic IntegrationFalse
PHI HandledTrue
Hardware DependencyFalse
Latency SensitiveUNKNOWN
Blast Radiusmedium
Migration RiskMEDIUM
Conf (Inferred)72%
Conf (Confirmed)55%
Verificationsingle_source
Cloud Connectivity
ComplianceHIPAA / HITRUST
Upstream (2)
Butterfly Network iQ Probe / Clinical Workstations: inferred: application_type=Cloud-Connected Gateway + vendor=Butterfly Network
HL7 Feed Source (EHR / Order Entry System): inferred: rvtools.vm_name='butterflyhl7'
Downstream (2)
Butterfly Cloud (vendor-hosted SaaS): confirmed: application_catalog.notes + rvtools.vm_notes
PACS / Radiology Information System (RIS): inferred: industry_context=healthcare + application_type=Cloud-Connected Gateway for ultrasound
Procurement Agent
SaaS Available—
SaaS VendorButterfly Network
SaaS ProductButterfly Cloud
Repurchase ViableINSUFFICIENT_EVIDENCE
BAA ConfirmedUNKNOWN
SaaS Confidence
Inferred Score72%
Confirmed Score0%
Verificationsingle_source
AWS Equiv.None
Provisioning Agent (Python engine — not LLM)
Current VMs3
Recommended VMs2
Right-Sizing25% reduction
Current TCO/yr$19,500
Projected Cloud TCO$10,400
Migration One-Time$2,080
Dual-Running Cost$9,967
-$2,947
Year-1 Net
$9,100
Steady-State
$15,253
3-Year
AWS TargetAmazon EC2
Azure TargetAzure Virtual Machines
Cost Modeldefault_multipliers
DEF: {'default_id': 'DEF-WL-001', 'field': 'workload_type', 'value': 'app_server', 'reason': 'workload_type indeterminable'}
DEF: {'default_id': 'DEF-MIG-001', 'field': 'migration_factor', 'value': 0.2, 'reason': 'migration_complexity not provided'}
DEF: {'default_id': 'DEF-COST-001', 'field': 'cost_model_basis', 'value': 'default_multipliers', 'reason': 'no customer cloud quote provided; using SRC-TCO-* and SRC-MULT-* registry defaults'}
HeartFlow Connect
Vendor: HeartFlow · 4 VMs · Criticality: High · HIPAA / HITRUST
REPURCHASE
40%
Confidence
28%
Data Quality
ROI: Year-1 Net: -$6,587 · Steady-State: $10,400/yr · 3-Year: $14,213
Overview
Backlog
Provenance
Technical Details
Agents
Recommendation
REPURCHASE ⏳ PENDING CONDITIONS
SRC-CAP-001 Binding cap · Ceiling: 40% · Threshold: 60%
Secondary: RETAIN
Confidence
40%
Viable, but specific gaps must close first.
ROI Summary (Python-computed, not LLM)
-$6,587
Year-1 Net
$10,400
Steady-State /yr
$14,213
3-Year Total
ROI variance: ±50%
⚙ Financials computed by provisioning_engine_v3 (Python) — not LLM-generated
In plain terms: migrating costs money up front — Year 1 is a net outflow of $6,587. From Year 2 onward the app saves $10,400 every year. The 3-Year total nets these against each other.
Finding
HeartFlow Connect is a cloud-connected DICOM gateway that already routes cardiac CT-FFR data to HeartFlow's vendor-hosted cloud platform, making full migration to HeartFlow Cloud the natural next step. However, the absence of a confirmed Business Associate Agreement for this PHI-handling workload is a compliance blocker that must be resolved before migration can proceed. Once BAA confirmation and contract terms are secured, this application is a strong REPURCHASE candidate that would eliminate four on-premises VMs and reduce annual infrastructure costs by approximately $10,400.
In plain terms: this app should move to the vendor's cloud product, but a few specific items must be confirmed first. See Conditions to Unlock below.
Conditions to Unlock (2 items)
This recommendation becomes executable once these conditions are confirmed:
☐ Obtain Signed Business Associate Agreement with HeartFlow to Unblock Migration Decision
Owner: Compliance Officer / Legal Counsel · Timeline: 21 days · Unlocks: compliance_close, cap_resolution, migration_commit
☐ Verify Whether HeartFlow Connect Is Used in Real-Time Clinical Decision Support
Owner: Cardiology Application Owner / Clinical Informatics · Timeline: 10 days · Unlocks: migration_commit, operational_safety
See Backlog tab for full action backlog and P2/P3 items.
This is the list of work needed to move this recommendation forward. The 2 P1 items are the priority — they unlock the recommendation. P2 and P3 items improve confidence and reduce risk but are not blockers. 'Story points' measure effort, not time or cost — see the Methodology tab for how they work.
40%
Current
60%
Threshold
55%
Ceiling
23
Story Points
Why This Score
Confidence is bound at 40 by SRC-CAP-001 (phi_handled=true with baa_confirmed=UNKNOWN), which is the single most restrictive of four firing caps; the pre-cap score of 63.5 reflects a genuinely strong REPURCHASE signal that is entirely blocked by the absence of a confirmed BAA. Resolving the BAA gap is the only action that lifts the binding cap and allows confidence to climb toward the SRC-CAP-002 ceiling of 55, at which point the remaining dimension-quality gaps become the next constraint.
Proceeding with REPURCHASE to HeartFlow Cloud without a confirmed BAA exposes the organization to a HIPAA Security Rule violation for PHI transmission to an unconfirmed business associate, and if life_safety_classification is subsequently confirmed true, the structural pre-check overrides REPURCHASE to RETAIN unconditionally.
Action Backlog — 6 Items · 23 Story Points · Ceiling: 55%
APP-023-001 Confirm Executed BAA with HeartFlow — Resolve SRC-CAP-001 Compliance Blocker
P1 BAA/Legal
Retrieve or execute a Business Associate Agreement (BAA) with HeartFlow that explicitly covers PHI transmission to HeartFlow Cloud (vendor-hosted SaaS). Confirm the BAA is current, executed by authorized signatories, and covers cloud-routed DICOM data. If an existing enterprise BAA is on file, confirm it extends to the HeartFlow Cloud SaaS delivery model. Provide the executed document or a complia
+15% confidence -0% ROI variance 👤 📋 · 5 pts
APP-023-002 Confirm Life Safety Classification with Cardiology Clinical Owner
P1 Security/Compliance
Engage the cardiology and clinical informatics application owner to determine whether HeartFlow Connect is used as a real-time decision-support tool in the cardiac catheterization lab or as an offline/batch CT-FFR analysis tool. If the application supports real-time intervention decisions (e.g., cath lab go/no-go for PCI), life_safety_classification must be set to true, which fires the structural
+0% confidence -0% ROI variance 👤 📋 · 3 pts
APP-023-003 Obtain HeartFlow Cloud SaaS Contract Terms and Migration Pricing
P2 Procurement
Engage HeartFlow account team to obtain: (1) customer-specific SaaS subscription pricing for HeartFlow Cloud, (2) contract terms including data portability, SLA, and termination provisions, (3) migration support scope (does HeartFlow provide a managed cutover from on-premises DICOM gateway to cloud-native connector?), and (4) confirmation of DICOM gateway architecture requirements — specifically w
+0% confidence -0% ROI variance 👤 📋 · 5 pts
APP-023-004 Produce Confirmed DICOM Dependency Map — Upstream Modalities and Downstream Consumers
P2 Dependency Mapping
Conduct a confirmed dependency mapping exercise to replace inferred upstream and downstream dependencies. Identify: (1) all upstream DICOM sources (CT scanner AE titles, PACS send rules, modality worklist integrations), (2) all downstream consumers of FFR results (cardiology reading workstations, PACS viewers, EHR result routing via HL7/FHIR or DICOM SR), and (3) any non-Epic EHR integration that
+0% confidence -0% ROI variance 👤 📋 · 5 pts
APP-023-005 Deploy 30-Day APM Telemetry Baseline on HeartFlow DICOM Gateway VMs
P2 Telemetry
Install lightweight APM agents (e.g., Datadog, New Relic, or Azure Monitor) on all four HeartFlow DICOM Virtual Appliance VMs to collect 30+ days of CPU, memory, disk I/O, and network throughput telemetry. This resolves the SRC-RS-002 hard rule cap (snapshot_count=1, apm_telemetry_available=false) and enables accurate right-sizing tier determination — potentially confirming ZOMBIE classification (
+0% confidence -0% ROI variance 👤 📋 · 3 pts
APP-023-006 Obtain HeartFlow Vendor Latency SLA and DICOM Routing Performance Requirements
P3 Vendor Engagement
Request HeartFlow technical specifications documenting DICOM routing latency requirements for the cloud-connected gateway architecture. Specifically: (1) is there a sub-second or sub-millisecond latency floor for DICOM series upload to HeartFlow Cloud?, (2) does HeartFlow Cloud support Direct Connect or ExpressRoute for HIPAA-compliant PHI transmission?, and (3) are there bandwidth minimums for la
+0% confidence -0% ROI variance 👤 📋 · 2 pts
This tab shows the evidence behind the recommendation. The seven dimensions are scored by quality — CONFIRMED evidence counts fully, INFERRED evidence counts less. The validator table confirms each analysis step passed its checks. Green means the step is sound.
Evidence Classification — 7 Dimensions
Dimension Classification Multiplier Contribution
Telemetry SINGLE 0.50 6.00
Dependency Mapping INFERRED 0.25 2.50
Procurement Contracts MISSING 0.00 0.00
Architecture Diagrams INFERRED 0.25 1.75
Security Compliance INFERRED 0.25 1.25
Business Criticality SINGLE 0.50 2.00
Dr Requirements MISSING 0.00 0.00
Applied Cap: SRC-CAP-001
Validation Reports — Per Agent
Agent Status Provenance Counts Defaults Used Rule Failures
telemetry PASS
SRC:2CUST:8COMP:5DEF:2
 None ✓ None
dependency PASS
SRC:6COMP:4DEF:4
 None ✓ None
procurement PASS
None ✓ None
provisioning PASS_WITH_DEFAULTS
SRC:7CUST:1COMP:9DEF:1
• {'default_id': 'DEF-MIG-001', 'field': 'migration_factor', 'value': 0.2, 'reason
 ✓ None
synthesizer PASS
None ✓ None
confidence_advisor PASS
COMP:15
 None ✓ None
This section contains the full technical reasoning behind the recommendation: the rules that fired, the decision path, the confidence math, and the open intelligence gaps. It is intended for analysts and engineers. Executive readers can rely on the Overview tab.
Rules Applied
These are the specific governance rules that produced this recommendation.
▸ RULE 1 — REPURCHASE
A vendor SaaS replacement is available and credible, and the app has no hardware constraint — so the recommendation is to replace it with the vendor's cloud service.
▸ SRC-CAP-001 (binding)
Confidence is capped at 40 because the app handles PHI and the vendor's Business Associate Agreement is not confirmed. PHI cannot move to a cloud vendor without a confirmed BAA.
▸ SRC-CAP-002
Confidence is capped at 55 because more than half of the seven evidence dimensions are inferred or missing — the conclusion is data-thin.
▸ SRC-CAP-003
Confidence is capped at 65 because none of the seven evidence dimensions reached CONFIRMED quality — there is no high-quality anchor for the analysis.
▸ SRC-CAP-004
Confidence is capped at 65 because telemetry is a single point-in-time snapshot with no ongoing monitoring — right-sizing needs usage patterns over time.
▸ RULE T2 — Right-Sizing Hard Cap
Right-sizing is limited to a moderate 25% reduction because telemetry is a single snapshot. More aggressive right-sizing needs 30+ days of monitoring data.
Decision Path
The step-by-step trace of how the recommendation was reached, in the system's own terms.
PRE-CHECK: hardware_dependency=false — structural blocker does not fire. RULE 1 (REPURCHASE): saas_alternative_available=true (SOURCED: application_catalog.notes); saas_confidence_inferred=72 >= 50; hardware_dependency=false — RULE 1 FIRES. strategic_recommendation=REPURCHASE. CONFIDENCE: base=50 + telemetry(SINGLE,12,0.50)=6.0 + dependency(INFERRED,10,0.25)=2.5 + procurement(MISSING,8,0.00)=0.0 + architecture(INFERRED,7,0.25)=1.75 + security(INFERRED,5,0.25)=1.25 + criticality(SINGLE,4,0.50)=2.0 + dr(MISSING,4,0.00)=0.0 = pre-cap 63.5. CAPS: SRC-CAP-001 fires (phi_handled=true, baa_confirmed=UNKNOWN) → ceiling 40; SRC-CAP-002 fires (5/7 INFERRED/MISSING) → ceiling 55; SRC-CAP-003 fires (zero CONFIRMED) → ceiling 65; SRC-CAP-004 fires (snapshot_count=1, apm=false) → ceiling 65. Binding cap: SRC-CAP-001 at 40. final_confidence=40. GATE (v4.1): final_confidence=40 < 60 AND strategic_recommendation=REPURCHASE (not RETAIN) → recommendation_status=PENDING_CONDITIONS; displayed_primary=REPURCHASE (gate does NOT force RETAIN in v4.1); confidence_gate_applied=true.
Cap Evaluation (all fired caps): SRC-CAP-001: PHI handled + BAA not confirmed → ceiling 40 (ceiling=40) ← BINDING SRC-CAP-002: ≥4 of 7 dimensions INFERRED/MISSING → ceiling 55 (ceiling=55) SRC-CAP-003: Zero CONFIRMED dimensions → ceiling 65 (ceiling=65) SRC-CAP-004: Single snapshot, no APM → ceiling 65 (ceiling=65)
Confidence Scoring Breakdown
How the confidence percentage was calculated, dimension by dimension.
Base score: 50 Evidence contributions: Telemetry SINGLE +6.00 Dependency Mapping INFERRED +2.50 Procurement Contracts MISSING +0.00 Architecture Diagrams INFERRED +1.75 Security Compliance INFERRED +1.25 Business Criticality SINGLE +2.00 Dr Requirements MISSING +0.00 Pre-cap score: 63.5 Binding cap applied: SRC-CAP-001 Final confidence: 40%
Intelligence Gaps
Specific things we don't yet know. Each gap is a reason the confidence score is held back.
BAA Confirmation — HeartFlow Cloud
phi_handled=true with baa_confirmed=UNKNOWN triggers SRC-CAP-001, binding confidence at 40 and placing recommendation in PENDING_CONDITIONS. BAA must be confirmed with HeartFlow before any PHI can be transmitted to vendor SaaS. This is the single highest-priority discovery item.
Contract Intelligence — SaaS Transition Terms
contract_intelligence=INSUFFICIENT_EVIDENCE; no data on SaaS pricing, migration support, transition timeline, or data portability. Cannot validate ROI projection or migration feasibility without contract review.
Latency Sensitivity Classification
latency_sensitive=UNKNOWN. HeartFlow CT-FFR workflow involves large DICOM series uploads; if latency requirements are confirmed, network architecture for SaaS migration must be validated. If latency_sensitive=true is confirmed, migration complexity increases.
Life Safety Classification
life_safety_classification=UNKNOWN. Cardiac CT-FFR is a clinical decision-support tool used in coronary artery disease diagnosis. If confirmed as life-safety, the structural pre-check fires and strategic_recommendation changes to RETAIN, overriding REPURCHASE.
APM Telemetry — Application Performance Baseline
apm_telemetry_available=false with snapshot_count=1. Right-sizing confidence is limited to RVTools memory snapshot (avg 4.5% active). SRC-RS-002 hard rule caps right-sizing at MODERATE. Full performance baseline needed to validate cloud instance sizing for SaaS migration cutover.
Technical Detail
A fuller technical explanation of the recommendation and its constraints.
Dependency agent confirms hardware_dependency=false and identifies HeartFlow Cloud as a confirmed downstream dependency, meaning the current on-prem gateway is already architecturally subordinate to the vendor SaaS platform. Procurement agent confirms saas_available=true with saas_confidence_inferred=72, satisfying Rule 1 threshold of 50. Telemetry agent flags CentOS 4/5/6 OS across all four VMs as an end-of-life risk that strengthens the case for eliminating the on-prem footprint. SRC-CAP-001 binds confidence at 40 due to phi_handled=true with baa_confirmed=UNKNOWN, placing the recommendation in PENDING_CONDITIONS status until BAA is confirmed.
Secondary Path
The fallback option if the primary recommendation cannot proceed.
Secondary Path — RETAIN
RETAIN is the secondary disposition if BAA cannot be confirmed or if life_safety_classification is subsequently confirmed true, which would trigger the structural pre-check and override REPURCHASE.
Right-Sizing Notes
Why the cloud sizing estimate is set where it is, based on the telemetry available.
RULE T2 TRIGGERED: snapshot_count=1 AND apm_telemetry_available=false. Fleet avg mem active pct = 4.5%, which under RULE T3 would classify as SRC-RS-004 (ZOMBIE). However, T2 hard rule unconditionally caps the maximum allowable tier at SRC-RS-002 (MODERATE, 0.25 reduction). Single-snapshot data cannot distinguish a genuinely idle VM from one sampled between burst cycles. REFUSAL: reasoning that 4.5% mem active 'clearly' indicates ZOMBIE was halted per T2 refusal trigger. All 4 VMs are HeartFlow DICOM virtual appliances. CPU utilization data is unavailable (cpu_avg_pct null for all 4 VMs). Additional telemetry (30+ days APM or multiple RVTools snapshots) is required before OVERSIZED or ZOMBIE classification can be applied. Application type is Cloud-Connected Gateway; RULE T1 life-safety override does not apply as life_safety_classification is absent from catalog.
This tab shows the raw output of each analysis agent — the specialized components that examine telemetry, dependencies, contracts, and cost. It is the most detailed and most technical view. Most readers will find the Overview and Backlog tabs sufficient.
Telemetry Agent
Utilization Profilemoderate
Snapshot Count1
APM AvailableFalse
Hard Rule TriggeredTrue
Right-Sizing Opp.True
Agent Confidence35%
Telemetry — VM Counts
Powered Off VMs0
Flagged VMs0
Mem Active % (avg)4.5%
Data WindowNone
Inventory Source
Dependency Agent
Epic IntegrationFalse
PHI HandledTrue
Hardware DependencyFalse
Latency SensitiveUNKNOWN
Blast Radiushigh
Migration RiskMEDIUM
Conf (Inferred)72%
Conf (Confirmed)58%
Verificationsingle_source
Cloud Connectivity
ComplianceHIPAA / HITRUST
Upstream (2)
PACS / Imaging Modality (CT Scanner): inferred: DICOM Virtual Appliance designation in all four vm_notes implies upstream DICOM source (CT/MR modality or…
Network / Firewall (MUSC-USB datacenter): confirmed: rvtools.datacenters = ['MUSC-USB']; all VMs co-located in single datacenter implying shared network fabric…
Downstream (3)
HeartFlow Cloud (vendor SaaS): confirmed: application_catalog.notes = 'SaaS alternative: HeartFlow Cloud (vendor-hosted SaaS)' — implies existing or…
Cardiology / Radiology Reading Workstations: inferred: DICOM gateway pattern implies downstream DICOM consumers (reading stations, cardiology PACS viewers) that…
EHR / Clinical Workflow (Epic or equivalent): inferred: High-criticality cardiac imaging result delivery in healthcare typically integrates with EHR for result…
Procurement Agent
SaaS Available—
SaaS VendorHeartFlow
SaaS ProductHeartFlow Cloud
Repurchase ViableINSUFFICIENT_EVIDENCE
BAA ConfirmedUNKNOWN
SaaS Confidence
Inferred Score72%
Confirmed Score0%
Verificationsingle_source
AWS Equiv.None
Provisioning Agent (Python engine — not LLM)
Current VMs4
Recommended VMs3
Right-Sizing25% reduction
Current TCO/yr$26,000
Projected Cloud TCO$15,600
Migration One-Time$3,120
Dual-Running Cost$13,867
-$6,587
Year-1 Net
$10,400
Steady-State
$14,213
3-Year
AWS TargetAmazon EC2
Azure TargetAzure Virtual Machines
Cost Modeldefault_multipliers
DEF: {'default_id': 'DEF-MIG-001', 'field': 'migration_factor', 'value': 0.2, 'reason': 'migration_complexity not provided'}
DEF: {'default_id': 'DEF-COST-001', 'field': 'cost_model_basis', 'value': 'default_multipliers', 'reason': 'no customer cloud quote provided; using SRC-TCO-* and SRC-MULT-* registry defaults'}
Vizai
Vendor: Viz.ai · 2 VMs · Criticality: High · HIPAA / HITRUST
REPURCHASE
55%
Confidence
32%
Data Quality
ROI: Year-1 Net: -$7,280 · Steady-State: $2,600/yr · 3-Year: -$2,080
Overview
Backlog
Provenance
Technical Details
Agents
Recommendation
REPURCHASE ⏳ PENDING CONDITIONS
SRC-CAP-002 Binding cap · Ceiling: 55% · Threshold: 60%
Secondary: RETAIN
Confidence
55%
Viable, but specific gaps must close first.
ROI Summary (Python-computed, not LLM)
-$7,280
Year-1 Net
$2,600
Steady-State /yr
-$2,080
3-Year Total
ROI variance: ±55%
⚙ Financials computed by provisioning_engine_v3 (Python) — not LLM-generated
In plain terms: migrating costs money up front — Year 1 is a net outflow of $7,280. From Year 2 onward the app saves $2,600 every year. The 3-Year total nets these against each other.
Finding
Viz.ai already offers a vendor-hosted SaaS product (Viz.ai Cloud) that would eliminate the two on-premises VMs currently running this AI stroke-detection workload. However, the BAA status is unconfirmed and procurement contract evidence is insufficient, so the migration cannot be executed until those compliance and commercial conditions are resolved. Once BAA confirmation and contract terms are validated, this application is a strong REPURCHASE candidate with an estimated $2,600 annual savings.
In plain terms: this app should move to the vendor's cloud product, but a few specific items must be confirmed first. See Conditions to Unlock below.
Conditions to Unlock (3 items)
This recommendation becomes executable once these conditions are confirmed:
☐ Verify Business Associate Agreement with Viz.ai for Cloud Migration
Owner: Legal / Compliance Officer · Timeline: 21 days · Unlocks: compliance_close, cap_resolution, migration_commit
☐ Secure Formal Contract and Pricing Proposal for Viz.ai Cloud
Owner: Procurement / Vendor Management · Timeline: 21 days · Unlocks: cost_clarity, cap_resolution, migration_commit
☐ Confirm Whether Patient Health Information Is Processed by Viz.ai
Owner: Data Owner / HIPAA Privacy Officer · Timeline: 7 days · Unlocks: compliance_close, cap_resolution
See Backlog tab for full action backlog and P2/P3 items.
This is the list of work needed to move this recommendation forward. The 3 P1 items are the priority — they unlock the recommendation. P2 and P3 items improve confidence and reduce risk but are not blockers. 'Story points' measure effort, not time or cost — see the Methodology tab for how they work.
55%
Current
60%
Threshold
62%
Ceiling
43
Story Points
Why This Score
The current confidence of 55 is hard-capped at the ceiling by SRC-CAP-002 (5 of 7 evidence dimensions are INFERRED or MISSING), which is the binding constraint — the pre-cap analytical score of 61.75 already exceeds the threshold, meaning the recommendation is analytically sound but evidence-deficient. No confidence delta is achievable until the binding cap is lifted by elevating the majority of dimensions to CONFIRMED through the backlog items above.
Executing REPURCHASE to Viz.ai Cloud without resolving BAA confirmation, procurement contract evidence, and clinical workflow continuity validation exposes the organization to HIPAA enforcement risk, unquantified SaaS transition costs, and potential patient safety impact from unvalidated SLA coverage for a life-safety stroke detection workload.
Action Backlog — 6 Items · 43 Story Points · Ceiling: 62%
APP-024-001 Confirm BAA with Viz.ai for Viz.ai Cloud SaaS Product
P1 BAA/Legal
Obtain and validate a signed Business Associate Agreement (BAA) with Viz.ai covering the Viz.ai Cloud SaaS product for the current healthcare entity. baa_confirmed=UNKNOWN because catalog.baa_valid=null. phi_handling=true and the application processes CT brain scan data for identified patients, constituting PHI under HIPAA. A confirmed BAA is a hard legal prerequisite before any PHI can be transmi
+0% confidence -0% ROI variance 👤 📋 · 8 pts
APP-024-002 Obtain Viz.ai Cloud Procurement Contract and SaaS Pricing Terms
P1 Procurement
Engage Viz.ai procurement to obtain a formal SaaS proposal or executed contract for Viz.ai Cloud. contract_intelligence=INSUFFICIENT_EVIDENCE: contract_acv=null, contract_renewal_date=null, and no RFP response or peer institution reference is on file. ROI projection carries 55% variance directly attributable to this gap. Collect: annual contract value (ACV), SaaS subscription pricing model, transi
+7% confidence -0% ROI variance 👤 📋 · 8 pts
APP-024-003 Confirm PHI Handling Classification and Catalog Field Normalization
P1 Security/Compliance
Resolve phi_handled=UNKNOWN caused by a catalog field name mismatch ('phi_handling' vs expected 'phi_handled'). The dependency agent assessed phi_risk as 'likely_phi' based on CT brain scan processing for identified patients, but the framework requires explicit confirmation. If phi_handled is confirmed true without a concurrent BAA confirmation, SRC-CAP-001 will fire and drop the confidence ceilin
+7% confidence -0% ROI variance 👤 📋 · 3 pts
APP-024-004 Collect Architecture Diagrams: PACS/CT Integration and Clinical Alert Routing
P2 Architecture
Architecture diagrams are MISSING for Viz.ai. The upstream PACS/CT scanner integration and downstream clinical alert routing (stroke team notification) are inferred only from VM notes. For a life-safety AI application (life_safety_classification=true), confirmed integration topology is required before SaaS cutover planning can begin. Collect: PACS system name and version, integration protocol (DIC
+7% confidence -0% ROI variance 👤 📋 · 8 pts
APP-024-005 Deploy 30-Day APM Telemetry Collection on VIZ-1 and VIZ-2
P2 Telemetry
apm_telemetry_available=false and snapshot_count=1 are the conditions triggering SRC-CAP-004 (ceiling 65) and the SRC-RS-002 hard cap (right-sizing capped at MODERATE). CPU utilization is entirely absent (cpu_avg_pct null on both VMs). A 30-day APM collection window on VIZ-1 and VIZ-2 will: (1) provide time-series CPU and memory data across varied clinical load periods including peak stroke alert
+0% confidence -0% ROI variance 👤 📋 · 13 pts
APP-024-006 Document DR Requirements and Validate Against Viz.ai Cloud SLA
P3 Architecture
dr_requirements=MISSING for Viz.ai. For a life-safety AI application (LVO stroke detection, life_safety_classification=true), RTO and RPO requirements must be formally documented and validated against the Viz.ai Cloud SaaS SLA before REPURCHASE can be executed. An unplanned outage of stroke detection AI during a high-volume period constitutes a patient safety risk. Collect: RTO/RPO requirements fr
+7% confidence -0% ROI variance 👤 📋 · 3 pts
This tab shows the evidence behind the recommendation. The seven dimensions are scored by quality — CONFIRMED evidence counts fully, INFERRED evidence counts less. The validator table confirms each analysis step passed its checks. Green means the step is sound.
Evidence Classification — 7 Dimensions
Dimension Classification Multiplier Contribution
Telemetry SINGLE 0.50 6.00
Dependency Mapping INFERRED 0.25 2.50
Procurement Contracts MISSING 0.00 0.00
Architecture Diagrams MISSING 0.00 0.00
Security Compliance INFERRED 0.25 1.25
Business Criticality SINGLE 0.50 2.00
Dr Requirements MISSING 0.00 0.00
Applied Cap: SRC-CAP-002
Validation Reports — Per Agent
Agent Status Provenance Counts Defaults Used Rule Failures
telemetry PASS
SRC:2CUST:9COMP:4DEF:2
 None ✓ None
dependency PASS
SRC:6CUST:3COMP:5DEF:2
 None ✓ None
procurement PASS
None ✓ None
provisioning PASS_WITH_DEFAULTS
SRC:7CUST:1COMP:9DEF:1
• {'default_id': 'DEF-MIG-001', 'field': 'migration_factor', 'value': 0.2, 'reason
 ✓ None
synthesizer PASS
None ✓ None
confidence_advisor PASS
COMP:15
 None ✓ None
This section contains the full technical reasoning behind the recommendation: the rules that fired, the decision path, the confidence math, and the open intelligence gaps. It is intended for analysts and engineers. Executive readers can rely on the Overview tab.
Rules Applied
These are the specific governance rules that produced this recommendation.
▸ RULE 1 — REPURCHASE
A vendor SaaS replacement is available and credible, and the app has no hardware constraint — so the recommendation is to replace it with the vendor's cloud service.
▸ SRC-CAP-002 (binding)
Confidence is capped at 55 because more than half of the seven evidence dimensions are inferred or missing — the conclusion is data-thin.
▸ SRC-CAP-003
Confidence is capped at 65 because none of the seven evidence dimensions reached CONFIRMED quality — there is no high-quality anchor for the analysis.
▸ SRC-CAP-004
Confidence is capped at 65 because telemetry is a single point-in-time snapshot with no ongoing monitoring — right-sizing needs usage patterns over time.
▸ RULE T2 — Right-Sizing Hard Cap
Right-sizing is limited to a moderate 25% reduction because telemetry is a single snapshot. More aggressive right-sizing needs 30+ days of monitoring data.
Decision Path
The step-by-step trace of how the recommendation was reached, in the system's own terms.
PRE-CHECK: hardware_dependency=false — structural blocker does not fire. RULE 1 (REPURCHASE): saas_alternative_available=true (Viz.ai Cloud per application notes and procurement output), saas_confidence_inferred=80 >= 50, hardware_dependency=false — Rule 1 FIRES. strategic_recommendation=REPURCHASE. CONFIDENCE SCORING: base=50, telemetry SINGLE (+6.0), dependency_mapping INFERRED (+2.5), procurement_contracts MISSING (+0.0), architecture_diagrams MISSING (+0.0), security_compliance INFERRED (+1.25), business_criticality SINGLE (+2.0), dr_requirements MISSING (+0.0) = pre-cap 61.75. CAP EVALUATION: SRC-CAP-001 does not fire (phi_handled=UNKNOWN, not confirmed true). SRC-CAP-002 fires (5/7 dimensions INFERRED/MISSING, cap=55). SRC-CAP-003 fires (zero CONFIRMED, cap=65). SRC-CAP-004 fires (snapshot_count=1, apm=false, cap=65). Lowest cap = SRC-CAP-002 at 55. final_confidence=55. GATE (v4.1): final_confidence=55 < 60 AND strategic_recommendation=REPURCHASE (not RETAIN) → recommendation_status=PENDING_CONDITIONS, confidence_gate_applied=true, displayed_primary=REPURCHASE (gate does NOT force RETAIN in v4.1). Conditions to unlock: (1) Confirm BAA with Viz.ai for SaaS product, (2) Obtain procurement contract terms for Viz.ai Cloud, (3) Validate clinical workflow continuity and latency SLA for stroke response use case, (4) Confirm phi_handled classification.
Cap Evaluation (all fired caps): SRC-CAP-002: ≥4 of 7 dimensions INFERRED/MISSING → ceiling 55 (ceiling=55) ← BINDING SRC-CAP-003: Zero CONFIRMED dimensions → ceiling 65 (ceiling=65) SRC-CAP-004: Single snapshot, no APM → ceiling 65 (ceiling=65)
Confidence Scoring Breakdown
How the confidence percentage was calculated, dimension by dimension.
Base score: 50 Evidence contributions: Telemetry SINGLE +6.00 Dependency Mapping INFERRED +2.50 Procurement Contracts MISSING +0.00 Architecture Diagrams MISSING +0.00 Security Compliance INFERRED +1.25 Business Criticality SINGLE +2.00 Dr Requirements MISSING +0.00 Pre-cap score: 61.75 Binding cap applied: SRC-CAP-002 Final confidence: 55%
Intelligence Gaps
Specific things we don't yet know. Each gap is a reason the confidence score is held back.
BAA Status Unconfirmed for Viz.ai Cloud SaaS
Cannot execute REPURCHASE to a vendor-hosted SaaS handling PHI without a confirmed BAA. phi_handled=UNKNOWN compounds this gap — if PHI is confirmed in scope, BAA is a hard legal prerequisite under HIPAA.
Procurement Contract Evidence Missing
contract_intelligence=INSUFFICIENT_EVIDENCE means licensing terms, SaaS pricing, and transition rights are unknown. ROI projection carries 55% variance until contract terms are confirmed.
PHI Handling Classification Unknown
phi_handled=UNKNOWN despite application processing CT brain scan data in a stroke detection workflow. If PHI is confirmed in scope, SRC-CAP-001 would apply, capping confidence at 40 until BAA is confirmed.
Clinical Workflow Continuity for Life-Safety Function
life_safety_classification=true — LVO stroke detection is time-critical. SaaS cutover requires validated latency SLA, uptime guarantees, and failover procedures before migration can be approved.
Architecture Diagrams Absent
No architecture artifacts available. PACS/CT scanner integration topology and downstream clinical alert routing are inferred only. Migration planning requires confirmed integration map.
DR Requirements Not Documented
No DR artifacts cited. For a life-safety AI application, RTO/RPO requirements must be validated against Viz.ai Cloud SaaS SLA before REPURCHASE can be executed.
APM Telemetry Unavailable / Single Snapshot
apm_telemetry_available=false and snapshot_count=1 limit right-sizing confidence. Right-sizing tier capped at MODERATE (SRC-RS-002) per hard rule.
Technical Detail
A fuller technical explanation of the recommendation and its constraints.
Procurement confirms Viz.ai Cloud SaaS is available with saas_confidence_inferred=80, and hardware_dependency=false removes the structural blocker for REPURCHASE. The application is explicitly typed as AI/Cloud-Connected and the vendor's own SaaS product is the natural target state, eliminating 2 on-premises VMs. However, baa_confirmed=UNKNOWN and contract_intelligence=INSUFFICIENT_EVIDENCE mean the compliance and commercial conditions for migration are unresolved. The life_safety_classification=true from the dependency agent (LVO stroke detection is a time-critical clinical function) elevates migration risk to HIGH and requires validated SaaS continuity planning before cutover.
Secondary Path
The fallback option if the primary recommendation cannot proceed.
Secondary Path — RETAIN
RETAIN is the secondary disposition if BAA cannot be confirmed or if clinical workflow validation reveals unacceptable latency or availability risk in the SaaS model for time-critical stroke response.
Right-Sizing Notes
Why the cloud sizing estimate is set where it is, based on the telemetry available.
Fleet avg mem active is 18.0% (VIZ-1: 16.0%, VIZ-2: 20.0%), which under Rule T3 would classify as OVERSIZED (SRC-RS-003, <25% threshold). However, Rule T2 is unconditionally triggered: snapshot_count=1 AND apm_telemetry_available=false. Per T2, SRC-RS-003 and SRC-RS-004 are FORBIDDEN; tier is hard-capped at SRC-RS-002 MODERATE (0.25 reduction). Single-snapshot data cannot distinguish a genuinely idle VM from one sampled between bursts — this reasoning is the explicit basis for the cap and is not overridable by utilization signal alone. CPU utilization data is entirely absent (cpu_avg_pct null on 2/2 VMs, no APM supplement). Rule T1 was evaluated: application_type='AI/Cloud-Connected' does not satisfy the Hardware-Integrated predicate, and life_safety_classification is UNKNOWN (absent from catalog); T1 does not fire. Right-sizing opportunity is affirmed at MODERATE tier pending deeper telemetry collection (30+ day APM or multi-snapshot RVTools).
This tab shows the raw output of each analysis agent — the specialized components that examine telemetry, dependencies, contracts, and cost. It is the most detailed and most technical view. Most readers will find the Overview and Backlog tabs sufficient.
Telemetry Agent
Utilization Profilemoderate
Snapshot Count1
APM AvailableFalse
Hard Rule TriggeredTrue
Right-Sizing Opp.True
Agent Confidence45%
Telemetry — VM Counts
Powered Off VMs0
Flagged VMs0
Mem Active % (avg)18.0%
Data WindowNone
Inventory Source
Dependency Agent
Epic IntegrationFalse
PHI HandledUNKNOWN
Hardware DependencyFalse
Latency SensitiveFalse
Blast Radiuscritical
Migration RiskHIGH
Conf (Inferred)78%
Conf (Confirmed)52%
Verificationsingle_source
Cloud Connectivity
ComplianceHIPAA / HITRUST
Upstream (2)
CT Scanner / PACS (Picture Archiving and Communication System): inferred: vm_notes state 'AI on CT brain scans' — CT imaging data must originate from a PACS or direct scanner feed
Hospital Network / WAN (MUSC-WAN): confirmed: rvtools_export.csv → datacenter = MUSC-WAN for both VMs
Downstream (2)
Stroke Response Clinical Workflow / Neurology Team Notification: inferred: LVO stroke detection AI implies downstream clinical alert/notification system (e.g., paging, EHR alert,…
EHR / Epic (potential): inferred: epic_integration=false per catalog, but clinical AI tools in stroke workflows commonly surface findings in…
Procurement Agent
SaaS Available—
SaaS VendorViz.ai
SaaS ProductViz.ai Cloud
Repurchase ViableINSUFFICIENT_EVIDENCE
BAA ConfirmedUNKNOWN
SaaS Confidence
Inferred Score80%
Confirmed Score0%
Verificationsingle_source
AWS Equiv.None
Provisioning Agent (Python engine — not LLM)
Current VMs2
Recommended VMs2
Right-Sizing25% reduction
Current TCO/yr$13,000
Projected Cloud TCO$10,400
Migration One-Time$2,080
Dual-Running Cost$7,800
-$7,280
Year-1 Net
$2,600
Steady-State
-$2,080
3-Year
AWS TargetAmazon EC2
Azure TargetAzure Virtual Machines
Cost Modeldefault_multipliers
DEF: {'default_id': 'DEF-MIG-001', 'field': 'migration_factor', 'value': 0.2, 'reason': 'migration_complexity not provided'}
DEF: {'default_id': 'DEF-COST-001', 'field': 'cost_model_basis', 'value': 'default_multipliers', 'reason': 'no customer cloud quote provided; using SRC-TCO-* and SRC-MULT-* registry defaults'}
GE Unified Viewer
Vendor: GE Healthcare · 9 VMs · Criticality: High · HIPAA / HITRUST — PHI handling status UN
REHOST
55%
Confidence
28%
Data Quality
ROI: Year-1 Net: -$16,813 · Steady-State: $22,100/yr · 3-Year: $27,387
Overview
Backlog
Provenance
Technical Details
Agents
Recommendation
REHOST ⏳ PENDING CONDITIONS
SRC-CAP-002 Binding cap · Ceiling: 55% · Threshold: 60%
Secondary: RETAIN
Confidence
55%
Viable, but specific gaps must close first.
ROI Summary (Python-computed, not LLM)
-$16,813
Year-1 Net
$22,100
Steady-State /yr
$27,387
3-Year Total
ROI variance: ±55%
⚙ Financials computed by provisioning_engine_v3 (Python) — not LLM-generated
In plain terms: migrating costs money up front — Year 1 is a net outflow of $16,813. From Year 2 onward the app saves $22,100 every year. The 3-Year total nets these against each other.
Finding
GE Unified Viewer is a 9-VM Windows Server 2019 imaging platform with no confirmed hardware dependency, making it a candidate for cloud rehosting. However, critical gaps in BAA status, procurement contracts, and architecture documentation prevent a fully executable migration plan at this time. Completing a BAA review, confirming cloud hosting eligibility with GE HealthCare, and mapping PACS/cardiology upstream dependencies are the priority actions to unlock this migration.
In plain terms: see the recommendation and conditions above.
Conditions to Unlock (3 items)
This recommendation becomes executable once these conditions are confirmed:
☐ Verify Business Associate Agreement Coverage for Cloud Migration
Owner: Compliance Officer / Legal · Timeline: 10 days · Unlocks: compliance_close, cap_resolution, migration_commit
☐ Confirm with GE HealthCare Whether Unified Viewer Can Be Hosted in the Cloud
Owner: Vendor Management / IT Procurement · Timeline: 14 days · Unlocks: migration_commit, cap_resolution, cost_clarity
☐ Collect 30 Days of Performance Data to Validate Right-Sizing Recommendation
Owner: Infrastructure / Cloud Engineering · Timeline: 30 days · Unlocks: cap_resolution, cost_clarity, operational_safety
See Backlog tab for full action backlog and P2/P3 items.
This is the list of work needed to move this recommendation forward. The 3 P1 items are the priority — they unlock the recommendation. P2 and P3 items improve confidence and reduce risk but are not blockers. 'Story points' measure effort, not time or cost — see the Methodology tab for how they work.
55%
Current
60%
Threshold
62%
Ceiling
18
Story Points
Why This Score
GE Unified Viewer is capped at 55 by SRC-CAP-002 because 5 of 7 evidence dimensions are INFERRED or MISSING — dependency mapping, procurement contracts, architecture diagrams, security/compliance, and DR requirements all lack confirmed sources. The pre-cap calculation of 61.75 reflects a structurally sound REHOST candidate, but the evidence base is too thin to support an executable migration decision without completing the items in this backlog.
Proceeding to migration planning without resolving BAA status, GE HealthCare cloud eligibility, and upstream PACS/CIS dependency confirmation exposes the organization to HIPAA compliance violations, vendor contract breach, and clinical imaging service disruption during a PHI-bearing cardiology and radiology workload migration.
Action Backlog — 6 Items · 18 Story Points · Ceiling: 62%
APP-025-001 Confirm BAA Status with GE HealthCare Technologies Inc. and Target Cloud Provider
P1 BAA/Legal
Obtain written confirmation that a valid BAA exists with GE HealthCare Technologies Inc. (post-January 2023 spin-off legal entity) covering cloud-hosted deployments of Unified Viewer. Separately, execute or confirm a BAA with the target cloud provider (AWS, Azure, GCP, or OCI). Any BAA executed prior to January 4, 2023 with 'GE Healthcare' as counterparty must be reviewed for novation status — the
+0% confidence -0% ROI variance 👤 📋 · 2 pts
APP-025-002 Engage GE HealthCare to Confirm Cloud Hosting Eligibility and Deployment Model
P1 Vendor Engagement
Submit a formal vendor inquiry to GE HealthCare Technologies Inc. to determine: (1) whether Unified Viewer is supported in customer-managed cloud environments (IaaS rehost on AWS/Azure/GCP/OCI), (2) whether GE HealthCare offers a cloud-hosted or SaaS-delivered version via the Edison platform or managed cloud model, (3) current license portability terms for cloud deployment, (4) any cloud-specific
+0% confidence -0% ROI variance 👤 📋 · 2 pts
APP-025-003 Deploy 30-Day APM Telemetry Collection for GE Unified Viewer Fleet
P1 Telemetry
Deploy application performance monitoring (APM) agents across all 9 GE Unified Viewer VMs to collect CPU utilization, memory active utilization, and network I/O time-series data over a minimum 30-day window. Current data is a single RVTools point-in-time snapshot with fleet_avg_mem_active_pct=12.6% and cpu_avg_pct=null for all 9 VMs. The SRC-RS-002 right-sizing recommendation (25% reduction, 9→7 V
+0% confidence -0% ROI variance 👤 📋 · 5 pts
APP-025-004 Confirm PACS/VNA and Cardiology Information System Upstream Dependencies
P2 Dependency Mapping
Conduct structured dependency interviews and network traffic analysis to confirm: (1) the upstream PACS or VNA system that feeds DICOM images to Unified Viewer — identify vendor, version, hosting model (on-prem vs. cloud), and migration readiness; (2) the Cardiology Information System (CIS) integrated with the 7 'GE UV Cardio Server' production nodes — identify vendor (e.g., GE Muse, Centricity),
+0% confidence -0% ROI variance 👤 📋 · 5 pts
APP-025-005 Retrieve and Review GE HealthCare Unified Viewer Procurement Contract
P2 Procurement
Retrieve the current GE HealthCare Unified Viewer support and licensing contract from the procurement system. Review for: (1) annual contract value (ACV) and renewal date, (2) cloud deployment rights or restrictions, (3) license portability clauses, (4) any existing cloud rider or amendment, (5) counterparty legal entity — confirm whether the contract names 'GE Healthcare' (pre-spin-off) or 'GE He
+7% confidence -0% ROI variance 👤 📋 · 2 pts
APP-025-006 Obtain or Reconstruct GE Unified Viewer Architecture Diagrams
P3 Architecture
Obtain existing architecture diagrams from GE HealthCare or the internal infrastructure team, or reconstruct from RVTools VM inventory and network scans. Document: (1) 9-VM cluster topology — admin node (UVPRODADMIN), 5 production nodes (UVPRODNODE1-5), test admin (UVTESTADMIN), test node (UVTESTNODE1), and the EAPROD cardiology node; (2) network segmentation and VLAN assignments; (3) shared stora
+7% confidence -0% ROI variance 👤 📋 · 2 pts
This tab shows the evidence behind the recommendation. The seven dimensions are scored by quality — CONFIRMED evidence counts fully, INFERRED evidence counts less. The validator table confirms each analysis step passed its checks. Green means the step is sound.
Evidence Classification — 7 Dimensions
Dimension Classification Multiplier Contribution
Telemetry SINGLE 0.50 6.00
Dependency Mapping INFERRED 0.25 2.50
Procurement Contracts MISSING 0.00 0.00
Architecture Diagrams MISSING 0.00 0.00
Security Compliance INFERRED 0.25 1.25
Business Criticality SINGLE 0.50 2.00
Dr Requirements MISSING 0.00 0.00
Applied Cap: SRC-CAP-002
Validation Reports — Per Agent
Agent Status Provenance Counts Defaults Used Rule Failures
telemetry PASS
SRC:2CUST:9COMP:3DEF:2
 None ✓ None
dependency PASS
SRC:2CUST:4COMP:3DEF:7
 None ✓ None
procurement PASS
None ✓ None
provisioning PASS_WITH_DEFAULTS
SRC:7CUST:1COMP:9DEF:1
• {'default_id': 'DEF-WL-001', 'field': 'workload_type', 'value': 'app_server', 'r
• {'default_id': 'DEF-MIG-001', 'field': 'migration_factor', 'value': 0.2, 'reason
 ✓ None
synthesizer PASS
None ✓ None
confidence_advisor PASS
COMP:15
 None ✓ None
This section contains the full technical reasoning behind the recommendation: the rules that fired, the decision path, the confidence math, and the open intelligence gaps. It is intended for analysts and engineers. Executive readers can rely on the Overview tab.
Rules Applied
These are the specific governance rules that produced this recommendation.
▸ RULE 6 — REHOST
No other qualifying path applies — the default for virtual workloads is to rehost to cloud IaaS.
▸ SRC-CAP-002 (binding)
Confidence is capped at 55 because more than half of the seven evidence dimensions are inferred or missing — the conclusion is data-thin.
▸ SRC-CAP-004
Confidence is capped at 65 because telemetry is a single point-in-time snapshot with no ongoing monitoring — right-sizing needs usage patterns over time.
▸ RULE T2 — Right-Sizing Hard Cap
Right-sizing is limited to a moderate 25% reduction because telemetry is a single snapshot. More aggressive right-sizing needs 30+ days of monitoring data.
Decision Path
The step-by-step trace of how the recommendation was reached, in the system's own terms.
PRE-CHECK: hardware_dependency=false, life_safety_classification=UNKNOWN — structural blocker does not fire. RULE 1 (REPURCHASE): saas_alternative_available=UNKNOWN, saas_confidence_inferred=30 < 50 threshold — does not fire. RULE 2 (RETIRE): no decommission annotations, no named successor — does not fire. RULE 3 (REFACTOR): vendor COTS (GE Healthcare), not internally developed — does not fire. RULE 4 (REHOST): VM-based=true (9 VMs), hardware_dependency=false, latency_sensitive=UNKNOWN (not confirmed true, not a structural blocker) — FIRES. strategic_recommendation=REHOST. CONFIDENCE GATE: pre_cap_total=61.75, SRC-CAP-002 applies (5/7 dimensions INFERRED/MISSING, cap=55), SRC-CAP-004 also applies (snapshot_count=1 AND apm_telemetry_available=false, cap=65) — lowest cap wins at 55. final_confidence=55 < 60 AND strategic_recommendation=REHOST (not RETAIN) → recommendation_status=PENDING_CONDITIONS, displayed_primary=REHOST (v4.1: gate does not force RETAIN). Conditions to unlock: (1) BAA confirmation with cloud provider, (2) GE HealthCare cloud hosting eligibility confirmation, (3) PACS/VNA upstream dependency mapping, (4) procurement contract review, (5) architecture diagram acquisition.
Cap Evaluation (all fired caps): SRC-CAP-002: ≥4 of 7 dimensions INFERRED/MISSING → ceiling 55 (ceiling=55) ← BINDING SRC-CAP-004: Single snapshot, no APM → ceiling 65 (ceiling=65)
Confidence Scoring Breakdown
How the confidence percentage was calculated, dimension by dimension.
Base score: 50 Evidence contributions: Telemetry SINGLE +6.00 Dependency Mapping INFERRED +2.50 Procurement Contracts MISSING +0.00 Architecture Diagrams MISSING +0.00 Security Compliance INFERRED +1.25 Business Criticality SINGLE +2.00 Dr Requirements MISSING +0.00 Pre-cap score: 61.75 Binding cap applied: SRC-CAP-002 Final confidence: 55%
Intelligence Gaps
Specific things we don't yet know. Each gap is a reason the confidence score is held back.
BAA Status Unconfirmed — PHI Handling Unknown
phi_handled=UNKNOWN and baa_confirmed=UNKNOWN. GE Unified Viewer processes DICOM imaging data which may contain PHI. Cloud migration cannot proceed without confirming PHI handling scope and executing a BAA with the target cloud provider. This is the binding compliance blocker.
GE HealthCare Cloud Hosting Eligibility Unknown
saas_available=UNKNOWN, saas_confidence_inferred=30. GE HealthCare may have cloud-hosted or hybrid deployment options for Unified Viewer, or may restrict deployment to customer-managed environments. Vendor confirmation is required before any migration planning can proceed.
PACS / VNA Upstream Dependency Unconfirmed
The PACS/imaging archive dependency is inferred from product class (DEFAULT_ASSUMPTION). If the upstream PACS is on-premises and cannot be co-migrated, Unified Viewer rehosting may introduce latency or connectivity constraints that affect clinical image rendering performance.
Cardiology Information System Integration Unconfirmed
All 7 production VMs are annotated as 'GE UV Cardio Server', indicating cardiology workflow integration. The upstream Cardiology Information System (CIS) is inferred only. If the CIS has hardware dependencies or cannot be migrated, this creates a co-dependency constraint on Unified Viewer migration.
No APM Telemetry — Single RVTools Snapshot
apm_telemetry_available=false and snapshot_count=1. Memory active utilization averages 12.6% but peak imaging load patterns (e.g., during radiology reading sessions) are unknown. Right-sizing to 7 VMs (SRC-RS-002) carries risk without longitudinal performance data.
Procurement Contracts — Insufficient Evidence
contract_intelligence=INSUFFICIENT_EVIDENCE. Current GE HealthCare contract terms, cloud rider availability, and license portability to cloud are unknown. Migration may require contract renegotiation or new licensing terms.
Architecture Diagrams Absent
No architecture diagrams available. The 9-VM cluster topology (admin node + 5 production nodes + test environment) is inferred from RVTools VM names and notes. Network segmentation, storage architecture, and HA/DR configuration are unknown.
DR Requirements Undefined
No DR requirements documented. As a High criticality imaging platform serving cardiology and radiology, RTO/RPO requirements are likely stringent but unconfirmed. Cloud DR architecture cannot be sized without this data.
Latency Sensitivity Classification Unknown
latency_sensitive=UNKNOWN. DICOM image rendering for clinical workflows may have latency requirements that affect cloud region selection and network architecture. Confirmation required before finalizing cloud deployment topology.
Technical Detail
A fuller technical explanation of the recommendation and its constraints.
GE Unified Viewer is a 9-VM Windows Server 2019 COTS imaging platform (GE Healthcare) with no confirmed hardware dependency, making REHOST the structurally appropriate disposition under Rule 4. The dependency agent identifies high blast radius and inferred PACS/cardiology upstream dependencies but confirms no hardware integration. The procurement agent returns saas_confidence_inferred=30 (below the 50 threshold for Rule 1) and UNKNOWN SaaS availability, ruling out REPURCHASE. No decommission annotations or named successors exist, ruling out RETIRE. The telemetry agent confirms a moderate utilization profile with right-sizing opportunity (SRC-RS-002, 25% reduction), but snapshot_count=1 and apm_telemetry_available=false limit confidence. SRC-CAP-002 applies (5 of 7 dimensions INFERRED or MISSING), capping confidence at 55 and triggering PENDING_CONDITIONS status.
Secondary Path
The fallback option if the primary recommendation cannot proceed.
Secondary Path — RETAIN
RETAIN is the secondary disposition if GE HealthCare confirms the application cannot be hosted outside a customer-managed on-premises environment, or if BAA and PHI handling review reveals cloud deployment is contractually prohibited. The $22,100 annual savings opportunity and right-sizing potential should be weighed against vendor cloud eligibility constraints during discovery.
Right-Sizing Notes
Why the cloud sizing estimate is set where it is, based on the telemetry available.
Fleet avg mem_active_pct = 12.6% across 9 powered-on VMs (range: 5.0%–21.0%). Under Rule T3 thresholds, 12.6% < 25% would classify as OVERSIZED (SRC-RS-003); however, Rule T2 (single-snapshot hard rule) is unconditionally triggered because snapshot_count=1 and apm_telemetry_available=false. Maximum allowable tier is capped at SRC-RS-002 (MODERATE, 0.25 reduction); SRC-RS-003 (OVERSIZED) and SRC-RS-004 (ZOMBIE) are forbidden. Single point-in-time memory readings cannot distinguish genuinely idle VMs from VMs sampled between burst cycles — this is the precise condition Rule T2 guards against. CPU utilization data is absent for all 9 VMs (cpu_avg_pct null throughout); no APM time-series available to supplement. Rule T1 life-safety override does not apply: application_type is Client/Server (not Hardware-Integrated) and life_safety_classification is absent from catalog (UNKNOWN). All 9 VMs run Windows Server 2019 (64-bit), which remains within extended support lifecycle through January 2029 — no EOL compliance flags generated. phi_handling=true is noted for downstream synthesizer; baa_valid=null will require SRC-CAP-001 evaluation at synthesis stage.
This tab shows the raw output of each analysis agent — the specialized components that examine telemetry, dependencies, contracts, and cost. It is the most detailed and most technical view. Most readers will find the Overview and Backlog tabs sufficient.
Telemetry Agent
Utilization Profilemoderate
Snapshot Count1
APM AvailableFalse
Hard Rule TriggeredTrue
Right-Sizing Opp.True
Agent Confidence45%
Telemetry — VM Counts
Powered Off VMs0
Flagged VMs0
Mem Active % (avg)12.6%
Data WindowNone
Inventory Source
Dependency Agent
Epic IntegrationFalse
PHI HandledUNKNOWN
Hardware DependencyFalse
Latency SensitiveUNKNOWN
Blast Radiushigh
Migration RiskMEDIUM
Conf (Inferred)62%
Conf (Confirmed)38%
Verificationsingle_source
Cloud Connectivity
ComplianceHIPAA / HITRUST
Upstream (3)
PACS / Imaging Archive (inferred): inferred: GE Unified Viewer is a DICOM viewer; requires upstream PACS or VNA image archive as structural dependency
Active Directory / Identity Provider: inferred: Windows Server 2019 cluster with admin node (UVPRODADMIN) implies domain-joined authentication dependency
Cardiology Information System (inferred): inferred: vm_notes across all production nodes reference 'GE UV Cardio Server', indicating cardiology workflow…
Downstream (2)
Clinical Workstations / Radiologists / Cardiologists (inferred): inferred: Unified Viewer is a clinical image rendering platform; downstream consumers are clinical end-users and…
Epic (confirmed absent): confirmed: catalog.epic_integration=false
Procurement Agent
SaaS Available—
SaaS VendorGE HealthCare
SaaS Product
Repurchase ViableUNKNOWN
BAA ConfirmedUNKNOWN
SaaS Confidence
Inferred Score30%
Confirmed Score0%
Verificationinferred_only
AWS Equiv.None
Provisioning Agent (Python engine — not LLM)
Current VMs9
Recommended VMs7
Right-Sizing25% reduction
Current TCO/yr$58,500
Projected Cloud TCO$36,400
Migration One-Time$7,280
Dual-Running Cost$31,633
-$16,813
Year-1 Net
$22,100
Steady-State
$27,387
3-Year
AWS TargetAmazon EC2
Azure TargetAzure Virtual Machines
Cost Modeldefault_multipliers
DEF: {'default_id': 'DEF-WL-001', 'field': 'workload_type', 'value': 'app_server', 'reason': 'workload_type indeterminable'}
DEF: {'default_id': 'DEF-MIG-001', 'field': 'migration_factor', 'value': 0.2, 'reason': 'migration_complexity not provided'}
DEF: {'default_id': 'DEF-COST-001', 'field': 'cost_model_basis', 'value': 'default_multipliers', 'reason': 'no customer cloud quote provided; using SRC-TCO-* and SRC-MULT-* registry defaults'}
LifeImage / Expedo
Vendor: NilRead (Hyland) · 4 VMs · Criticality: High · HIPAA / HITRUST
REPURCHASE
55%
Confidence
32%
Data Quality
ROI: Year-1 Net: -$6,587 · Steady-State: $10,400/yr · 3-Year: $14,213
Overview
Backlog
Provenance
Technical Details
Agents
Recommendation
REPURCHASE ⏳ PENDING CONDITIONS
SRC-CAP-002 Binding cap · Ceiling: 55% · Threshold: 60%
Secondary: REHOST
Confidence
55%
Viable, but specific gaps must close first.
ROI Summary (Python-computed, not LLM)
-$6,587
Year-1 Net
$10,400
Steady-State /yr
$14,213
3-Year Total
ROI variance: ±60%
⚙ Financials computed by provisioning_engine_v3 (Python) — not LLM-generated
In plain terms: migrating costs money up front — Year 1 is a net outflow of $6,587. From Year 2 onward the app saves $10,400 every year. The 3-Year total nets these against each other.
Finding
LifeImage / Expedo (NilRead) is a strong REPURCHASE candidate: the same vendor, Hyland, offers a cloud-hosted version of NilRead that would eliminate four aging on-premises VMs — including an end-of-life Windows Server 2012 R2 and SQL 2012 stack. Before committing to the cloud transition, MUSC must confirm a Business Associate Agreement with Hyland for the cloud offering and validate the DICOM/image-exchange architecture to ensure no hidden hardware or latency constraints exist. Resolving these two conditions would unlock an estimated $10,400 in annual savings with a 60% variance range reflecting current evidence gaps.
In plain terms: this app should move to the vendor's cloud product, but a few specific items must be confirmed first. See Conditions to Unlock below.
Conditions to Unlock (2 items)
This recommendation becomes executable once these conditions are confirmed:
☐ Verify Business Associate Agreement with Hyland for Cloud Migration
Owner: Compliance Officer / Procurement Lead · Timeline: 14 days · Unlocks: compliance_close, cap_resolution, migration_commit
☐ Document DICOM Image Routing and Integration Architecture
Owner: Enterprise Architect / Radiology Informatics Lead · Timeline: 21 days · Unlocks: migration_commit, cap_resolution, operational_safety
See Backlog tab for full action backlog and P2/P3 items.
This is the list of work needed to move this recommendation forward. The 2 P1 items are the priority — they unlock the recommendation. P2 and P3 items improve confidence and reduce risk but are not blockers. 'Story points' measure effort, not time or cost — see the Methodology tab for how they work.
55%
Current
60%
Threshold
64%
Ceiling
42
Story Points
Why This Score
The confidence score of 55 is the result of a pre-cap calculation of 63.5 being reduced by the binding SRC-CAP-002 ceiling (5 of 7 evidence dimensions are INFERRED or MISSING: dependency mapping, procurement contracts, architecture diagrams, security/compliance, and DR requirements), with SRC-CAP-004 also applied but non-binding at 65. The score cannot exceed 55 until the majority of INFERRED/MISSING dimensions are promoted to CONFIRMED through the discovery activities in this backlog.
Committing to REPURCHASE before BAA confirmation and architecture validation risks an active HIPAA compliance violation and potential migration failure if undiscovered DICOM routing constraints, latency requirements, or Hyland post-acquisition BAA gaps cannot be resolved within the discovery window.
Action Backlog — 6 Items · 42 Story Points · Ceiling: 64%
APP-026-001 Confirm BAA with Hyland for NilRead Cloud (Post-Acquisition Novation)
P1 BAA/Legal
Engage MUSC compliance officer and Hyland account team to confirm whether a valid Business Associate Agreement exists under the Hyland entity following its acquisition of LifeImage/NilRead. The original LifeImage BAA may not automatically extend to Hyland without explicit novation. Scope must cover the cloud-hosted NilRead offering specifically. This is the single most critical blocker: phi_handli
+0% confidence -0% ROI variance 👤 📋 · 5 pts
APP-026-002 Validate DICOM Routing and Image Exchange Architecture Diagrams
P1 Architecture
Produce current-state architecture diagrams covering: (1) DICOM source routing from modalities and PACS to LifeImage/Expedo, (2) HL7 interface connections to downstream clinical systems, (3) external image exchange partner integrations via the Expedo requestor module, (4) cross-datacenter topology between MUSC-ART and MUSC-USB (active-active vs. active-passive vs. test/prod split), and (5) any lat
+9% confidence -0% ROI variance 👤 📋 · 8 pts
APP-026-003 Deploy APM Telemetry Across LifeImage/Expedo Fleet (30-Day Collection)
P2 Telemetry
Deploy application performance monitoring (APM) tooling across all four LifeImage/Expedo VMs spanning MUSC-ART and MUSC-USB datacenters. Collect a minimum 30-day continuous telemetry window capturing CPU utilization (currently null across all 4 VMs), memory active percentage, disk I/O, and network throughput — including DICOM image transfer volumes. This resolves SRC-CAP-004 (snapshot_count=1, apm
+0% confidence -0% ROI variance 👤 📋 · 13 pts
APP-026-004 Document DR/RTO/RPO Requirements for Radiology Image Exchange Platform
P2 Architecture
Obtain formal DR requirements documentation from the Radiology Informatics team and MUSC IT leadership specifying RTO and RPO targets for the LifeImage/Expedo platform. The application is classified High criticality and handles PHI across two datacenters, strongly implying DR requirements exist — but no documentation has been cited. DR requirements are currently MISSING, contributing a 10% penalty
+9% confidence -0% ROI variance 👤 📋 · 3 pts
APP-026-005 Engage Hyland Procurement for NilRead Cloud Contract and ACV Intelligence
P2 Procurement
Initiate formal procurement engagement with Hyland to obtain: (1) ACV and renewal terms for the existing on-premises NilRead/LifeImage license, (2) customer-specific RFP response for NilRead Cloud migration from MUSC's multi-DC on-premises deployment, (3) Expedo requestor workflow compatibility confirmation in the cloud-hosted model, and (4) cloud transition rights under the current license agreem
+9% confidence -0% ROI variance 👤 📋 · 5 pts
APP-026-006 EOL OS Risk Assessment and Remediation Roadmap for LifeImage/Expedo Fleet
P3 Security/Compliance
Conduct a formal EOL OS risk assessment covering all four LifeImage/Expedo VMs: lifeimageexpedo (Windows Server 2012 R2 + SQL 2012, EOL Oct 2023), lifeimagelink-v (CentOS 4/5/6/7, EOL range 2012–2024), lifeimageprod (Linux kernel 3.x, EOL 2015, distribution unidentified), and lifeimagetest (RHEL 6, EOL Nov 2020). All four VMs are running end-of-life operating systems on a PHI-handling platform, co
+9% confidence -0% ROI variance 👤 📋 · 8 pts
This tab shows the evidence behind the recommendation. The seven dimensions are scored by quality — CONFIRMED evidence counts fully, INFERRED evidence counts less. The validator table confirms each analysis step passed its checks. Green means the step is sound.
Evidence Classification — 7 Dimensions
Dimension Classification Multiplier Contribution
Telemetry SINGLE 0.50 6.00
Dependency Mapping INFERRED 0.25 2.50
Procurement Contracts MISSING 0.00 0.00
Architecture Diagrams INFERRED 0.25 1.75
Security Compliance INFERRED 0.25 1.25
Business Criticality SINGLE 0.50 2.00
Dr Requirements MISSING 0.00 0.00
Applied Cap: SRC-CAP-002
Validation Reports — Per Agent
Agent Status Provenance Counts Defaults Used Rule Failures
telemetry PASS
SRC:2CUST:13COMP:5DEF:1
 None ✓ None
dependency PASS
SRC:8COMP:4DEF:6
 None ✓ None
procurement PASS
None ✓ None
provisioning PASS_WITH_DEFAULTS
SRC:7CUST:1COMP:9DEF:1
• {'default_id': 'DEF-WL-001', 'field': 'workload_type', 'value': 'app_server', 'r
• {'default_id': 'DEF-MIG-001', 'field': 'migration_factor', 'value': 0.2, 'reason
 ✓ None
synthesizer PASS
None ✓ None
confidence_advisor PASS
COMP:16
 None ✓ None
This section contains the full technical reasoning behind the recommendation: the rules that fired, the decision path, the confidence math, and the open intelligence gaps. It is intended for analysts and engineers. Executive readers can rely on the Overview tab.
Rules Applied
These are the specific governance rules that produced this recommendation.
▸ RULE 1 — REPURCHASE
A vendor SaaS replacement is available and credible, and the app has no hardware constraint — so the recommendation is to replace it with the vendor's cloud service.
▸ SRC-CAP-002 (binding)
Confidence is capped at 55 because more than half of the seven evidence dimensions are inferred or missing — the conclusion is data-thin.
▸ SRC-CAP-004
Confidence is capped at 65 because telemetry is a single point-in-time snapshot with no ongoing monitoring — right-sizing needs usage patterns over time.
▸ RULE T2 — Right-Sizing Hard Cap
Right-sizing is limited to a moderate 25% reduction because telemetry is a single snapshot. More aggressive right-sizing needs 30+ days of monitoring data.
Decision Path
The step-by-step trace of how the recommendation was reached, in the system's own terms.
PRE-CHECK: hardware_dependency=false, life_safety_classification=false → pre-check does not fire. RULE 1: saas_alternative_available=true (Hyland NilRead Cloud), saas_confidence_inferred=60 ≥ 50, hardware_dependency=false → RULE 1 FIRES → strategic_recommendation=REPURCHASE. CONFIDENCE: base=50 + telemetry(SINGLE,12,0.50)=6.0 + dependency(INFERRED,10,0.25)=2.5 + procurement(MISSING,8,0.00)=0.0 + architecture(INFERRED,7,0.25)=1.75 + security(INFERRED,5,0.25)=1.25 + criticality(SINGLE,4,0.50)=2.0 + dr(MISSING,4,0.00)=0.0 = pre-cap 63.5. CAP EVALUATION: SRC-CAP-001 not applicable (phi_handled=UNKNOWN, not confirmed true). SRC-CAP-002: 5/7 dimensions INFERRED or MISSING → cap 55 APPLIES. SRC-CAP-003: zero CONFIRMED → cap 65 (not binding). SRC-CAP-004: snapshot_count=1 AND apm=false → cap 65 (not binding). Lowest cap = 55. final_confidence=55. GATE: 55 < 60 AND strategic_recommendation=REPURCHASE → recommendation_status=PENDING_CONDITIONS, confidence_gate_applied=true. displayed_primary=REPURCHASE (v4.1: gate does not force RETAIN). ROI VARIANCE: base=5 + telemetry_penalty=20 (apm unavailable) + procurement_penalty=15 (MISSING) + arch_penalty=10 (INFERRED) + dep_penalty=10 (INFERRED) + dr_penalty=10 (MISSING) = 70 → capped at 60%.
Cap Evaluation (all fired caps): SRC-CAP-002: ≥4 of 7 dimensions INFERRED/MISSING → ceiling 55 (ceiling=55) ← BINDING SRC-CAP-004: Single snapshot, no APM → ceiling 65 (ceiling=65)
Confidence Scoring Breakdown
How the confidence percentage was calculated, dimension by dimension.
Base score: 50 Evidence contributions: Telemetry SINGLE +6.00 Dependency Mapping INFERRED +2.50 Procurement Contracts MISSING +0.00 Architecture Diagrams INFERRED +1.75 Security Compliance INFERRED +1.25 Business Criticality SINGLE +2.00 Dr Requirements MISSING +0.00 Pre-cap score: 63.5 Binding cap applied: SRC-CAP-002 Final confidence: 55%
Intelligence Gaps
Specific things we don't yet know. Each gap is a reason the confidence score is held back.
BAA Confirmation — Hyland Cloud Offering
phi_handling=true in application catalog; baa_confirmed=UNKNOWN for Hyland NilRead Cloud. Without a confirmed BAA, HIPAA-compliant cloud transition cannot proceed. This is the binding condition for REPURCHASE execution.
PHI Handling Confirmation
phi_handled=UNKNOWN in dependency agent (catalog reports true but no dependency-layer confirmation). If confirmed true, SRC-CAP-001 would apply (cap 40) unless BAA is simultaneously confirmed, further constraining confidence.
APM / Application Performance Telemetry
apm_telemetry_available=false; right-sizing estimate (SRC-RS-002, 25% reduction) and cloud performance baseline are estimate-only. Actual workload sizing for NilRead Cloud cannot be validated without APM data.
Architecture Diagrams — DICOM Routing and Image Exchange Topology
No architecture diagrams available. DICOM source routing, HL7 interface connections, and external image exchange partner integrations are inferred from vendor domain knowledge. Hidden latency or integration constraints may exist.
EOL OS and Database Stack
lifeimageexpedo VM runs Windows Server 2012 R2 and SQL 2012, both past end of extended support. Security vulnerability exposure is active; migration timeline should be treated as urgent regardless of 6R disposition.
DR Requirements Documentation
No DR documentation cited. RTO/RPO for radiology image exchange platform is unknown. High criticality designation implies DR requirements exist but are unvalidated.
Contract Intelligence — Hyland Licensing
contract_intelligence=INSUFFICIENT_EVIDENCE. Existing on-prem license terms, cloud transition rights, and SaaS pricing are unknown. Procurement negotiation scope cannot be estimated.
Technical Detail
A fuller technical explanation of the recommendation and its constraints.
Procurement agent confirms Hyland offers NilRead Cloud / Hyland Medical Imaging as a cloud-hosted SaaS equivalent (saas_confidence_inferred=60, meeting the Rule 1 threshold of ≥50), and dependency agent confirms hardware_dependency=false, clearing the absolute REPURCHASE blocker. Rule 1 fires: REPURCHASE is the strategic recommendation. Confidence is capped at 55 by SRC-CAP-002 (5 of 7 dimensions are INFERRED or MISSING) and SRC-CAP-004 (snapshot_count=1, apm_telemetry_available=false), placing the recommendation in PENDING_CONDITIONS status. The binding conditions are BAA confirmation for the Hyland cloud offering (phi_handling=true in catalog, baa_confirmed=UNKNOWN) and architecture validation of DICOM routing and image exchange topology. The EOL OS/DB stack (Windows Server 2012 R2 + SQL 2012 on lifeimageexpedo) adds urgency to the migration timeline.
Secondary Path
The fallback option if the primary recommendation cannot proceed.
Secondary Path — REHOST
If REPURCHASE conditions cannot be met within the discovery window, REHOST to IaaS (AWS/Azure) is the fallback — the workload is VM-based with no hardware dependency, making lift-and-shift structurally viable, though it would not resolve the EOL OS/DB risk without remediation.
Right-Sizing Notes
Why the cloud sizing estimate is set where it is, based on the telemetry available.
Single point-in-time RVTools snapshot with no APM telemetry. Hard rule T2 triggered: snapshot_count=1 AND apm_telemetry_available=false. Maximum allowed tier is SRC-RS-002 (MODERATE, 0.25 reduction); SRC-RS-003 (OVERSIZED) and SRC-RS-004 (ZOMBIE) are forbidden. Fleet avg mem active pct is 25.8%, which independently falls in the MODERATE band (25-60%) under Rule T3, consistent with the hard-rule cap. CPU utilization data is unavailable for all 4 VMs (cpu_avg_pct null across fleet). Application type is Client/Server with no life_safety_classification present; Rule T1 override does not fire. Four VMs carry EOL operating systems on a PHI-handling application, generating active HIPAA Security Rule compliance flags.
This tab shows the raw output of each analysis agent — the specialized components that examine telemetry, dependencies, contracts, and cost. It is the most detailed and most technical view. Most readers will find the Overview and Backlog tabs sufficient.
Telemetry Agent
Utilization Profilemoderate
Snapshot Count1
APM AvailableFalse
Hard Rule TriggeredTrue
Right-Sizing Opp.True
Agent Confidence45%
Telemetry — VM Counts
Powered Off VMs0
Flagged VMs0
Mem Active % (avg)25.8%
Data WindowNone
Inventory Source
Dependency Agent
Epic IntegrationFalse
PHI HandledUNKNOWN
Hardware DependencyFalse
Latency SensitiveUNKNOWN
Blast Radiushigh
Migration RiskMEDIUM
Conf (Inferred)72%
Conf (Confirmed)48%
Verificationsingle_source
Cloud Connectivity
ComplianceHIPAA / HITRUST
Upstream (3)
Radiology PACS / Modality Worklist: inferred: LifeImage/Expedo is a medical image exchange platform; upstream DICOM sources (modalities, PACS) are…
SQL Server 2012 (embedded, lifeimageexpedo VM): confirmed: rvtools.lifeimageexpedo.note — 'Windows Server 2012 R2 w/SQL 2012'
Active Directory / Identity Provider: inferred: Windows Server 2012 R2 environment; Client/Server application type implies domain authentication
Downstream (3)
Radiology Informatics Team Workflows (MUSC): confirmed: rvtools.lifeimageprod.note — 'Radiology Informatics for Lifeimage Server Upgrades'; rvtools.lifeimagelink-v.n…
External Image Exchange Partners (referring providers, external health systems): inferred: LifeImage product function is inter-institutional image exchange; 'Expedo' is the LifeImage…
Clinician Workstations / NilRead Viewer Clients: inferred: NilRead (Hyland) is a zero-footprint DICOM viewer; downstream consumers are radiologists and referring…
Procurement Agent
SaaS Available—
SaaS VendorHyland
SaaS ProductNilRead Cloud / Hyland Medical Imaging (cloud-hosted NilRead)
Repurchase ViableINSUFFICIENT_EVIDENCE
BAA ConfirmedUNKNOWN
SaaS Confidence
Inferred Score60%
Confirmed Score0%
Verificationinferred_only
AWS Equiv.None
Provisioning Agent (Python engine — not LLM)
Current VMs4
Recommended VMs3
Right-Sizing25% reduction
Current TCO/yr$26,000
Projected Cloud TCO$15,600
Migration One-Time$3,120
Dual-Running Cost$13,867
-$6,587
Year-1 Net
$10,400
Steady-State
$14,213
3-Year
AWS TargetAmazon EC2
Azure TargetAzure Virtual Machines
Cost Modeldefault_multipliers
DEF: {'default_id': 'DEF-WL-001', 'field': 'workload_type', 'value': 'app_server', 'reason': 'workload_type indeterminable'}
DEF: {'default_id': 'DEF-MIG-001', 'field': 'migration_factor', 'value': 0.2, 'reason': 'migration_complexity not provided'}
DEF: {'default_id': 'DEF-COST-001', 'field': 'cost_model_basis', 'value': 'default_multipliers', 'reason': 'no customer cloud quote provided; using SRC-TCO-* and SRC-MULT-* registry defaults'}
Magview
Vendor: Magview · 5 VMs · Criticality: High · HIPAA / HITRUST
REPURCHASE
40%
Confidence
42%
Data Quality
ROI: Year-1 Net: -$10,227 · Steady-State: $11,700/yr · 3-Year: $13,173
Overview
Backlog
Provenance
Technical Details
Agents
Recommendation
REPURCHASE ⏳ PENDING CONDITIONS
SRC-CAP-001 Binding cap · Ceiling: 40% · Threshold: 60%
Secondary: REHOST
Confidence
40%
Viable, but specific gaps must close first.
ROI Summary (Python-computed, not LLM)
-$10,227
Year-1 Net
$11,700
Steady-State /yr
$13,173
3-Year Total
ROI variance: ±50%
⚙ Financials computed by provisioning_engine_v3 (Python) — not LLM-generated
In plain terms: migrating costs money up front — Year 1 is a net outflow of $10,227. From Year 2 onward the app saves $11,700 every year. The 3-Year total nets these against each other.
Finding
Magview's own vendor offers a cloud-hosted SaaS version (Magview Cloud) that would eliminate all 5 on-premises VMs and transfer operational responsibility to the vendor, with an estimated $11,700 in annual savings. However, because Magview handles mammography data containing protected health information, a Business Associate Agreement with the SaaS vendor must be confirmed before any migration can proceed. Once BAA status and contract terms are verified, this is a strong candidate for a direct SaaS transition with minimal re-engineering.
In plain terms: this app should move to the vendor's cloud product, but a few specific items must be confirmed first. See Conditions to Unlock below.
Conditions to Unlock (3 items)
This recommendation becomes executable once these conditions are confirmed:
☐ Obtain Business Associate Agreement from Magview for Cloud Product
Owner: Compliance Officer / Legal Counsel · Timeline: 10 days · Unlocks: compliance_close, cap_resolution, migration_commit
☐ Collect SaaS Pricing and Contract Details from Magview Vendor
Owner: Procurement / Vendor Management · Timeline: 14 days · Unlocks: cost_clarity, migration_commit
☐ Confirm That Magview Cloud Supports the Existing MyChart Integration
Owner: Integration Architect / Vendor Engagement Lead · Timeline: 21 days · Unlocks: migration_commit, operational_safety
See Backlog tab for full action backlog and P2/P3 items.
This is the list of work needed to move this recommendation forward. The 3 P1 items are the priority — they unlock the recommendation. P2 and P3 items improve confidence and reduce risk but are not blockers. 'Story points' measure effort, not time or cost — see the Methodology tab for how they work.
40%
Current
60%
Threshold
65%
Ceiling
24
Story Points
Why This Score
Confidence is hard-capped at 40 by SRC-CAP-001 because phi_handled=true and baa_confirmed=UNKNOWN — the BAA status for Magview Cloud cannot be inferred from a null catalog field under v3.1 tri-state rules, and PHI migration without a confirmed BAA is a HIPAA compliance blocker regardless of technical readiness. Secondary caps SRC-CAP-003 (zero CONFIRMED dimensions) and SRC-CAP-004 (single snapshot, no APM) would bind at 65 once SRC-CAP-001 is resolved, meaning the BAA item alone is insufficient to reach the 60-point execution threshold — telemetry and at least one additional dimension confirmation are also required.
Proceeding with Magview Cloud migration without a confirmed BAA exposes the organization to HIPAA enforcement risk and potential OCR investigation; the REPURCHASE recommendation remains PENDING_CONDITIONS until BAA confirmation, contract terms, and MyChart interface compatibility are all resolved.
Action Backlog — 6 Items · 24 Story Points · Ceiling: 65%
APP-027-001 Confirm BAA with Magview Cloud SaaS Vendor (SRC-CAP-001 Resolution)
P1 BAA/Legal
Obtain and execute a Business Associate Agreement (BAA) with Magview covering the Magview Cloud SaaS product. The current catalog field baa_valid=null maps to UNKNOWN under v3.1 tri-state rules, triggering SRC-CAP-001 which caps confidence at 40 and blocks EXECUTABLE status. The compliance officer or legal team must request BAA documentation from Magview, confirm it covers PHI processed in the clo
+25% confidence -0% ROI variance 👤 📋 · 3 pts
APP-027-002 Obtain Magview Cloud Contract Terms, ACV, and Renewal Date
P1 Procurement
Engage Magview's sales or account management team to obtain customer-specific SaaS subscription pricing (ACV), contract renewal date, and termination/transition clauses for the existing on-premises license. Current contract_intelligence=INSUFFICIENT_EVIDENCE and contract_acv=null, which applies a 15% procurement penalty to ROI variance. Procurement must request: (1) Magview Cloud subscription quot
+0% confidence -0% ROI variance 👤 📋 · 3 pts
APP-027-003 Validate Magview Cloud Epic/MyChart Interface Compatibility
P1 Vendor Engagement
The on-premises DMZ VM (magvwebdmz) provides a confirmed MyChart/Epic patient portal interface. It is currently unknown whether Magview Cloud SaaS supports an equivalent integration path. This must be validated before the DMZ VM can be decommissioned. Engage Magview's technical team to confirm: (1) whether Magview Cloud supports HL7/FHIR or proprietary MyChart interfacing, (2) what network connect
+0% confidence -0% ROI variance 👤 📋 · 5 pts
APP-027-004 Identify and Confirm Upstream PACS / Modality Worklist Integration
P2 Dependency Mapping
Magview is a mammography tracking platform that structurally requires an upstream imaging data feed — either a PACS system or a modality worklist — but no specific upstream system is named in the application catalog or VM notes. This dependency is currently flagged as DEFAULT_ASSUMPTION (DEF-003). The dependency mapping team must interview the application owner (Wendy Ketchum per VM notes) and rad
+0% confidence -0% ROI variance 👤 📋 · 5 pts
APP-027-005 Deploy Extended Telemetry Collection Across All 5 Magview VMs
P2 Telemetry
Current telemetry is a single RVTools point-in-time snapshot with no APM data, triggering SRC-CAP-004 and the TELEMETRY-DEPTH-INSUFFICIENT compliance flag. The right-sizing recommendation (25% reduction, SRC-RS-002 MODERATE) is capped at MODERATE precisely because single-snapshot data cannot distinguish genuinely idle VMs from those sampled between bursts. Deploy a 30-day continuous monitoring sol
+0% confidence -0% ROI variance 👤 📋 · 5 pts
APP-027-006 Document Architecture Diagrams and DR / RTO-RPO Requirements
P2 Architecture
No formal architecture documentation exists for Magview; application topology is inferred from VM notes only (architecture_diagrams=INFERRED). DR requirements are entirely absent (dr_requirements=MISSING), which applies a 10% DR penalty to ROI variance. For a High-criticality PHI system, RTO/RPO requirements must be defined and validated against Magview Cloud SLA before migration can be approved.
+0% confidence -0% ROI variance 👤 📋 · 3 pts
This tab shows the evidence behind the recommendation. The seven dimensions are scored by quality — CONFIRMED evidence counts fully, INFERRED evidence counts less. The validator table confirms each analysis step passed its checks. Green means the step is sound.
Evidence Classification — 7 Dimensions
Dimension Classification Multiplier Contribution
Telemetry SINGLE 0.50 6.00
Dependency Mapping SINGLE 0.50 5.00
Procurement Contracts INFERRED 0.25 2.00
Architecture Diagrams INFERRED 0.25 1.75
Security Compliance INFERRED 0.25 1.25
Business Criticality SINGLE 0.50 2.00
Dr Requirements MISSING 0.00 0.00
Applied Cap: SRC-CAP-001
Validation Reports — Per Agent
Agent Status Provenance Counts Defaults Used Rule Failures
telemetry PASS
SRC:2CUST:9COMP:4DEF:2
 None ✓ None
dependency PASS
SRC:9COMP:5DEF:3
 None ✓ None
procurement PASS
None ✓ None
provisioning PASS_WITH_DEFAULTS
SRC:7CUST:1COMP:9DEF:1
• {'default_id': 'DEF-WL-001', 'field': 'workload_type', 'value': 'app_server', 'r
• {'default_id': 'DEF-MIG-001', 'field': 'migration_factor', 'value': 0.2, 'reason
 ✓ None
synthesizer PASS
None ✓ None
confidence_advisor PASS
COMP:27
 None ✓ None
This section contains the full technical reasoning behind the recommendation: the rules that fired, the decision path, the confidence math, and the open intelligence gaps. It is intended for analysts and engineers. Executive readers can rely on the Overview tab.
Rules Applied
These are the specific governance rules that produced this recommendation.
▸ RULE 1 — REPURCHASE
A vendor SaaS replacement is available and credible, and the app has no hardware constraint — so the recommendation is to replace it with the vendor's cloud service.
▸ SRC-CAP-001 (binding)
Confidence is capped at 40 because the app handles PHI and the vendor's Business Associate Agreement is not confirmed. PHI cannot move to a cloud vendor without a confirmed BAA.
▸ SRC-CAP-003
Confidence is capped at 65 because none of the seven evidence dimensions reached CONFIRMED quality — there is no high-quality anchor for the analysis.
▸ SRC-CAP-004
Confidence is capped at 65 because telemetry is a single point-in-time snapshot with no ongoing monitoring — right-sizing needs usage patterns over time.
▸ RULE T2 — Right-Sizing Hard Cap
Right-sizing is limited to a moderate 25% reduction because telemetry is a single snapshot. More aggressive right-sizing needs 30+ days of monitoring data.
Decision Path
The step-by-step trace of how the recommendation was reached, in the system's own terms.
PRE-CHECK: hardware_dependency=false, life_safety_classification=false — structural blocker does not fire. RULE 1 (REPURCHASE): saas_available=true (source: application notes + procurement output), saas_confidence_inferred=72 >= 50, hardware_dependency=false — RULE 1 FIRES. strategic_recommendation=REPURCHASE. CONFIDENCE SCORING: base=50, telemetry(SINGLE)=+6.0, dependency_mapping(SINGLE)=+5.0, procurement_contracts(INFERRED)=+2.0, architecture_diagrams(INFERRED)=+1.75, security_compliance(INFERRED)=+1.25, business_criticality(SINGLE)=+2.0, dr_requirements(MISSING)=+0.0; pre_cap_total=68.0. CAP EVALUATION: SRC-CAP-001 fires (phi_handled=true AND baa_confirmed=UNKNOWN != true) → cap 40; SRC-CAP-003 fires (zero CONFIRMED dimensions) → cap 65; SRC-CAP-004 fires (snapshot_count=1 AND apm_telemetry_available=false) → cap 65. Binding cap: SRC-CAP-001 at 40. final_confidence=40. GATE (v4.1): final_confidence=40 < 60 AND strategic_recommendation=REPURCHASE (not RETAIN) → recommendation_status=PENDING_CONDITIONS, displayed_primary=REPURCHASE (gate does NOT force RETAIN in v4.1), confidence_gate_applied=true.
Cap Evaluation (all fired caps): SRC-CAP-001: PHI handled + BAA not confirmed → ceiling 40 (ceiling=40) ← BINDING SRC-CAP-003: Zero CONFIRMED dimensions → ceiling 65 (ceiling=65) SRC-CAP-004: Single snapshot, no APM → ceiling 65 (ceiling=65)
Confidence Scoring Breakdown
How the confidence percentage was calculated, dimension by dimension.
Base score: 50 Evidence contributions: Telemetry SINGLE +6.00 Dependency Mapping SINGLE +5.00 Procurement Contracts INFERRED +2.00 Architecture Diagrams INFERRED +1.75 Security Compliance INFERRED +1.25 Business Criticality SINGLE +2.00 Dr Requirements MISSING +0.00 Pre-cap score: 68.0 Binding cap applied: SRC-CAP-001 Final confidence: 40%
Intelligence Gaps
Specific things we don't yet know. Each gap is a reason the confidence score is held back.
BAA Confirmation with Magview Cloud (P1 — Binding Cap Trigger)
SRC-CAP-001 caps confidence at 40 because phi_handled=true and baa_confirmed=UNKNOWN. A confirmed BAA with Magview Cloud is the single highest-priority action to unlock EXECUTABLE status. Without it, PHI migration to SaaS cannot proceed under HIPAA.
Contract Intelligence — Existing License Terms and SaaS Pricing
contract_intelligence=INSUFFICIENT_EVIDENCE means ROI projection carries 15% procurement penalty in variance calculation. Current contract renewal dates, termination clauses, and Magview Cloud subscription pricing are unknown, making the $11,700 savings estimate unreliable.
Upstream PACS / Modality Worklist Integration Confirmation
Upstream imaging data feed (PACS or modality worklist) is inferred as structurally required for mammography tracking but not explicitly named in catalog or VM notes. SaaS migration requires a validated integration path for this upstream dependency to avoid clinical workflow disruption.
MyChart / Epic Integration Compatibility with Magview Cloud
The DMZ server (magvwebdmz) provides confirmed MyChart integration. It is unknown whether Magview Cloud SaaS supports equivalent Epic/MyChart interfacing. This must be validated before decommissioning the on-premises DMZ VM.
DR / RTO-RPO Requirements
No DR documentation exists for this application. As a High-criticality system handling PHI, RTO/RPO requirements must be defined and validated against Magview Cloud SLA before migration.
Architecture Diagrams
No formal architecture documentation cited. Application topology inferred from VM notes only. Formal diagrams required to confirm all integration points before SaaS cutover planning.
Technical Detail
A fuller technical explanation of the recommendation and its constraints.
Rule 1 (REPURCHASE) fires: saas_available=true (Magview Cloud confirmed in application notes and procurement output), saas_confidence_inferred=72 exceeds the 50 threshold, and hardware_dependency=false removes the absolute blocker. The pre-check does not apply as neither hardware_dependency nor life_safety_classification is true. SRC-CAP-001 applies because phi_handled=true and baa_confirmed=UNKNOWN, capping confidence at 40 and triggering PENDING_CONDITIONS status. The secondary recommendation is REHOST as a fallback if SaaS transition is blocked by BAA or contract constraints, given the VM-based architecture with no hardware dependency.
Secondary Path
The fallback option if the primary recommendation cannot proceed.
Secondary Path — REHOST
REHOST is viable as a fallback: all 5 VMs run Windows Server 2022, no hardware dependency exists, and the workload is standard client/server architecture. A lift-and-shift to IaaS would preserve the existing MyChart DMZ integration while eliminating on-premises infrastructure costs, though it would not capture the full operational savings of SaaS.
Right-Sizing Notes
Why the cloud sizing estimate is set where it is, based on the telemetry available.
RULE T1 does not apply: application_type is Client/Server, not Hardware-Integrated. RULE T2 fires: snapshot_count=1 AND apm_telemetry_available=false; hard_rule_triggered=true; maximum allowed tier is SRC-RS-002 (MODERATE, 0.25 reduction). Fleet avg mem_active_pct=11.8% would indicate OVERSIZED under RULE T3, but single-snapshot data cannot safely distinguish a genuinely idle VM from one sampled between bursts; OVERSIZED (SRC-RS-003) and ZOMBIE (SRC-RS-004) are forbidden. All 5 VMs run Windows Server 2022 (64-bit), which is not EOL; no HIPAA-EOL compliance flags generated. CPU utilization data is unavailable: cpu_avg_pct is null on all 5 VMs and no APM telemetry is present.
This tab shows the raw output of each analysis agent — the specialized components that examine telemetry, dependencies, contracts, and cost. It is the most detailed and most technical view. Most readers will find the Overview and Backlog tabs sufficient.
Telemetry Agent
Utilization Profilemoderate
Snapshot Count1
APM AvailableFalse
Hard Rule TriggeredTrue
Right-Sizing Opp.True
Agent Confidence45%
Telemetry — VM Counts
Powered Off VMs0
Flagged VMs0
Mem Active % (avg)11.8%
Data WindowNone
Inventory Source
Dependency Agent
Epic IntegrationFalse
PHI HandledTrue
Hardware DependencyFalse
Latency SensitiveFalse
Blast Radiushigh
Migration RiskMEDIUM
Conf (Inferred)72%
Conf (Confirmed)58%
Verificationsingle_source
Cloud Connectivity
ComplianceHIPAA / HITRUST
Upstream (2)
MyChart / Epic Patient Portal: confirmed: rvtools.vm_notes (magvwebdmz)
Radiology / Mammography Imaging Source (PACS or modality worklist): inferred: Magview is a mammography tracking platform; upstream imaging data feed is structurally required but not…
Downstream (3)
MyChart (Epic Patient Portal) — results delivery: confirmed: rvtools.vm_notes (magvwebdmz)
Clinical End Users (Radiologists, Ordering Clinicians): confirmed: rvtools.vm_notes (magvweb)
Reporting / Analytics consumers of Magview DB: inferred: magvdb is a dedicated production database server; downstream reporting consumers are structurally expected…
Procurement Agent
SaaS Available—
SaaS VendorMagview
SaaS ProductMagview Cloud
Repurchase ViableINSUFFICIENT_EVIDENCE
BAA ConfirmedUNKNOWN
SaaS Confidence
Inferred Score72%
Confirmed Score0%
Verificationsingle_source
AWS Equiv.None
Provisioning Agent (Python engine — not LLM)
Current VMs5
Recommended VMs4
Right-Sizing25% reduction
Current TCO/yr$32,500
Projected Cloud TCO$20,800
Migration One-Time$4,160
Dual-Running Cost$17,767
-$10,227
Year-1 Net
$11,700
Steady-State
$13,173
3-Year
AWS TargetAmazon EC2
Azure TargetAzure Virtual Machines
Cost Modeldefault_multipliers
DEF: {'default_id': 'DEF-WL-001', 'field': 'workload_type', 'value': 'app_server', 'reason': 'workload_type indeterminable'}
DEF: {'default_id': 'DEF-MIG-001', 'field': 'migration_factor', 'value': 0.2, 'reason': 'migration_complexity not provided'}
DEF: {'default_id': 'DEF-COST-001', 'field': 'cost_model_basis', 'value': 'default_multipliers', 'reason': 'no customer cloud quote provided; using SRC-TCO-* and SRC-MULT-* registry defaults'}
PenRad
Vendor: PenRad Technologies · 2 VMs · Criticality: High · HIPAA / HITRUST
REPURCHASE
40%
Confidence
38%
Data Quality
ROI: Year-1 Net: -$7,280 · Steady-State: $2,600/yr · 3-Year: -$2,080
Overview
Backlog
Provenance
Technical Details
Agents
Recommendation
REPURCHASE ⏳ PENDING CONDITIONS
SRC-CAP-001 Binding cap · Ceiling: 40% · Threshold: 60%
Secondary: RETAIN (pending BAA confirmation)
Confidence
40%
Viable, but specific gaps must close first.
ROI Summary (Python-computed, not LLM)
-$7,280
Year-1 Net
$2,600
Steady-State /yr
-$2,080
3-Year Total
ROI variance: ±40%
⚙ Financials computed by provisioning_engine_v3 (Python) — not LLM-generated
In plain terms: migrating costs money up front — Year 1 is a net outflow of $7,280. From Year 2 onward the app saves $2,600 every year. The 3-Year total nets these against each other.
Finding
PenRad is a mammography reporting platform handling protected health information, and the same vendor offers a cloud-hosted version (PenRad Cloud) that could eliminate the two on-premises VMs. However, before any migration can proceed, MUSC must confirm that a valid Business Associate Agreement is in place with PenRad Technologies for the cloud product — without this, HIPAA compliance cannot be assured. Once the BAA is confirmed and the SaaS contract terms are validated, this application is a strong candidate for a same-vendor cloud migration.
In plain terms: this app should move to the vendor's cloud product, but a few specific items must be confirmed first. See Conditions to Unlock below.
Conditions to Unlock (2 items)
This recommendation becomes executable once these conditions are confirmed:
☐ Verify Business Associate Agreement with PenRad Technologies
Owner: MUSC Legal / Compliance Officer · Timeline: 21 days · Unlocks: compliance_close, cap_resolution, migration_commit
☐ Get PenRad Cloud Pricing, SLA, and Migration Terms from Vendor
Owner: Procurement / Vendor Management · Timeline: 30 days · Unlocks: cost_clarity, migration_commit
See Backlog tab for full action backlog and P2/P3 items.
This is the list of work needed to move this recommendation forward. The 2 P1 items are the priority — they unlock the recommendation. P2 and P3 items improve confidence and reduce risk but are not blockers. 'Story points' measure effort, not time or cost — see the Methodology tab for how they work.
40%
Current
60%
Threshold
65%
Ceiling
32
Story Points
Why This Score
PenRad carries a final confidence of 40 because SRC-CAP-001 is the binding cap: phi_handled=true and baa_confirmed=UNKNOWN (catalog.baa_valid=null) unconditionally impose a ceiling of 40, overriding the pre-cap score of 68.0. No backlog item can raise confidence above 40 in this run because the ceiling equals the current score; all confidence_delta values are set to zero per the B2 ceiling-bound rule, and the backlog is oriented entirely toward cap resolution and ROI variance reduction.
Proceeding with REPURCHASE to PenRad Cloud without BAA confirmation exposes MUSC to a HIPAA compliance violation for PHI transmitted to a vendor SaaS environment without a valid Business Associate Agreement.
Action Backlog — 6 Items · 32 Story Points · Ceiling: 65%
APP-028-001 Confirm BAA with PenRad Technologies for PenRad Cloud (SRC-CAP-001 resolution)
P1 BAA/Legal
Retrieve or execute a Business Associate Agreement (BAA) with PenRad Technologies covering MUSC as a covered entity for the PenRad Cloud SaaS product. The catalog field baa_valid is null, making baa_confirmed=UNKNOWN. Because phi_handled=true and baa_confirmed is not true, SRC-CAP-001 is the binding confidence cap at ceiling 40 — the lowest cap in the stack. No migration activity can proceed until
+25% confidence -0% ROI variance 👤 📋 · 3 pts
APP-028-002 Obtain PenRad Cloud SaaS contract terms, pricing, and migration proposal
P1 Procurement
Engage PenRad Technologies procurement to obtain: (1) customer-specific SaaS migration proposal for PenRad Cloud including ACV, per-seat or per-study pricing, and contract term options; (2) SLA commitments covering uptime, RTO/RPO, and data residency; (3) data migration scope and timeline; (4) subprocessor and hosting environment disclosure for HIPAA compliance review. Current contract_intelligenc
+0% confidence -0% ROI variance 👤 📋 · 8 pts
APP-028-003 Confirm PACS/DICOM upstream integration pathway for PenRad Cloud compatibility
P2 Dependency Mapping
Identify and document the upstream PACS or radiology image archive that feeds DICOM studies into PenRad. The dependency is currently inferred from industry pattern (mammography reporting systems structurally require a PACS upstream) but is not confirmed from any catalog or network artifact. Confirm: (1) PACS system name and version; (2) DICOM integration method (DICOM C-STORE, DICOM Q/R, or HL7 OR
+0% confidence -0% ROI variance 👤 📋 · 5 pts
APP-028-004 Confirm HL7/ADT feed source and downstream result routing for PenRad
P2 Dependency Mapping
Identify and document: (1) the upstream HL7/ADT feed delivering patient demographics and orders to PenRad — confirm whether this originates from Epic, a standalone interface engine, or another HIS; (2) the downstream result routing path — confirm whether PenRad sends results to Epic or another EMR via HL7 outbound, direct DB link, or interface engine (catalog.epic_integration=false does not rule o
+0% confidence -0% ROI variance 👤 📋 · 5 pts
APP-028-005 Deploy 30-day APM telemetry collection for PenRad fleet (penrad-v, penradtest-v)
P2 Telemetry
Install APM or extended monitoring on penrad-v and penradtest-v to collect a minimum 30-day CPU and RAM time-series. Current data is a single point-in-time RVTools snapshot (snapshot_count=1, apm_telemetry_available=false), which triggers SRC-CAP-004 (ceiling 65) and the SRC-RS-002 hard cap (MODERATE right-sizing only). Fleet avg_mem_active_pct=2.5% cannot be confirmed as representative of steady-
+0% confidence -0% ROI variance 👤 📋 · 8 pts
APP-028-006 Define and document DR/RTO-RPO requirements for PenRad mammography reporting
P3 Architecture
DR requirements are MISSING across all agent outputs (dr_requirements classification=MISSING, contribution=0.0). For a High-criticality mammography reporting system handling PHI, RTO and RPO must be formally defined and validated against PenRad Cloud SLA commitments before migration can proceed. Engage the MUSC Radiology department and IT leadership to document: (1) maximum tolerable downtime (RTO
+0% confidence -0% ROI variance 👤 📋 · 3 pts
This tab shows the evidence behind the recommendation. The seven dimensions are scored by quality — CONFIRMED evidence counts fully, INFERRED evidence counts less. The validator table confirms each analysis step passed its checks. Green means the step is sound.
Evidence Classification — 7 Dimensions
Dimension Classification Multiplier Contribution
Telemetry SINGLE 0.50 6.00
Dependency Mapping SINGLE 0.50 5.00
Procurement Contracts INFERRED 0.25 2.00
Architecture Diagrams INFERRED 0.25 1.75
Security Compliance INFERRED 0.25 1.25
Business Criticality SINGLE 0.50 2.00
Dr Requirements MISSING 0.00 0.00
Applied Cap: SRC-CAP-001
Validation Reports — Per Agent
Agent Status Provenance Counts Defaults Used Rule Failures
telemetry PASS
SRC:2CUST:10COMP:3DEF:2
 None ✓ None
dependency PASS
SRC:2CUST:4COMP:5DEF:1
 None ✓ None
procurement PASS
None ✓ None
provisioning PASS_WITH_DEFAULTS
SRC:7CUST:1COMP:9DEF:1
• {'default_id': 'DEF-WL-001', 'field': 'workload_type', 'value': 'app_server', 'r
• {'default_id': 'DEF-MIG-001', 'field': 'migration_factor', 'value': 0.2, 'reason
 ✓ None
synthesizer PASS
None ✓ None
confidence_advisor PASS
COMP:37
 None ✓ None
This section contains the full technical reasoning behind the recommendation: the rules that fired, the decision path, the confidence math, and the open intelligence gaps. It is intended for analysts and engineers. Executive readers can rely on the Overview tab.
Rules Applied
These are the specific governance rules that produced this recommendation.
▸ RULE 1 — REPURCHASE
A vendor SaaS replacement is available and credible, and the app has no hardware constraint — so the recommendation is to replace it with the vendor's cloud service.
▸ SRC-CAP-001 (binding)
Confidence is capped at 40 because the app handles PHI and the vendor's Business Associate Agreement is not confirmed. PHI cannot move to a cloud vendor without a confirmed BAA.
▸ SRC-CAP-003
Confidence is capped at 65 because none of the seven evidence dimensions reached CONFIRMED quality — there is no high-quality anchor for the analysis.
▸ SRC-CAP-004
Confidence is capped at 65 because telemetry is a single point-in-time snapshot with no ongoing monitoring — right-sizing needs usage patterns over time.
▸ RULE T2 — Right-Sizing Hard Cap
Right-sizing is limited to a moderate 25% reduction because telemetry is a single snapshot. More aggressive right-sizing needs 30+ days of monitoring data.
Decision Path
The step-by-step trace of how the recommendation was reached, in the system's own terms.
PRE-CHECK: hardware_dependency=false, life_safety_classification=false → pre-check does not fire. RULE 1 (REPURCHASE): saas_alternative_available=true (procurement.saas_available=true, notes confirm PenRad cloud), saas_confidence_inferred=70 >= 50, hardware_dependency=false → Rule 1 FIRES. strategic_recommendation=REPURCHASE. CONFIDENCE: base=50 + telemetry(SINGLE,12,0.50)=6.0 + dependency(SINGLE,10,0.50)=5.0 + procurement(INFERRED,8,0.25)=2.0 + arch(INFERRED,7,0.25)=1.75 + security(INFERRED,5,0.25)=1.25 + criticality(SINGLE,4,0.50)=2.0 + dr(MISSING,4,0.00)=0.0 → pre-cap=68.0. CAP EVALUATION: SRC-CAP-001 fires (phi_handled=true, baa_confirmed=UNKNOWN≠true) → ceiling 40; SRC-CAP-003 fires (zero CONFIRMED dimensions) → ceiling 65; SRC-CAP-004 fires (snapshot_count=1, apm_telemetry_available=false) → ceiling 65. Binding cap = SRC-CAP-001 at 40. final_confidence=40. GATE (v4.1): final_confidence=40 < 60 AND strategic_recommendation=REPURCHASE (not RETAIN) → recommendation_status=PENDING_CONDITIONS, confidence_gate_applied=true, displayed_primary=REPURCHASE (gate does NOT force RETAIN in v4.1).
Cap Evaluation (all fired caps): SRC-CAP-001: PHI handled + BAA not confirmed → ceiling 40 (ceiling=40) ← BINDING SRC-CAP-003: Zero CONFIRMED dimensions → ceiling 65 (ceiling=65) SRC-CAP-004: Single snapshot, no APM → ceiling 65 (ceiling=65)
Confidence Scoring Breakdown
How the confidence percentage was calculated, dimension by dimension.
Base score: 50 Evidence contributions: Telemetry SINGLE +6.00 Dependency Mapping SINGLE +5.00 Procurement Contracts INFERRED +2.00 Architecture Diagrams INFERRED +1.75 Security Compliance INFERRED +1.25 Business Criticality SINGLE +2.00 Dr Requirements MISSING +0.00 Pre-cap score: 68.0 Binding cap applied: SRC-CAP-001 Final confidence: 40%
Intelligence Gaps
Specific things we don't yet know. Each gap is a reason the confidence score is held back.
BAA Confirmation for PenRad Cloud
SRC-CAP-001 binding cap (ceiling 40) cannot be lifted until baa_confirmed=true. HIPAA compliance for PHI in SaaS environment is unverified. This is the single highest-priority unlock condition for REPURCHASE execution.
Contract Intelligence for PenRad Cloud SaaS
contract_intelligence=INSUFFICIENT_EVIDENCE and saas_confidence_confirmed=0. Pricing, SLA, data migration terms, and transition timeline for PenRad Cloud are unknown. Procurement validation required before migration commitment.
PACS / DICOM Integration Confirmation
Upstream PACS dependency is inferred from industry pattern, not confirmed. PenRad Cloud must support the same DICOM integration pathway as the on-prem deployment. Failure to confirm this could block clinical workflow continuity post-migration.
HL7 / ADT Feed Integration Confirmation
HL7/ADT upstream dependency is inferred. PenRad Cloud's compatibility with MUSC's existing HL7 interface engine must be validated to ensure patient demographic and order data continuity.
APM Telemetry and Performance Baseline
apm_telemetry_available=false and snapshot_count=1. No application-layer performance baseline exists. Right-sizing for cloud and SLA definition for PenRad Cloud cannot be validated without this data. SRC-CAP-004 ceiling (65) applies.
DR / RTO-RPO Requirements
No DR requirements data present across any agent output. For a High-criticality mammography reporting system handling PHI, RTO/RPO must be defined and confirmed against PenRad Cloud SLA before migration.
Radiologist Workstation Client Compatibility
Downstream radiologist workstation clients are inferred from Client/Server application type. Compatibility of existing client software with PenRad Cloud must be confirmed to avoid clinical workflow disruption.
Technical Detail
A fuller technical explanation of the recommendation and its constraints.
Procurement agent confirmed saas_available=true with saas_confidence_inferred=70 for PenRad Cloud (same vendor), satisfying Rule 1 REPURCHASE criteria; dependency agent confirmed hardware_dependency=false and latency_sensitive=false, removing all structural blockers. The pre-check does not fire (no hardware + life-safety combination). However, SRC-CAP-001 applies because phi_handled=true and baa_confirmed=UNKNOWN, capping confidence at 40 and triggering PENDING_CONDITIONS status. The strategic disposition is REPURCHASE; execution requires BAA confirmation and contract validation before migration can proceed.
Secondary Path
The fallback option if the primary recommendation cannot proceed.
Secondary Path — RETAIN (pending BAA confirmation)
If BAA cannot be confirmed for PenRad Cloud, REHOST to IaaS (AWS EC2 / Azure VM) is a viable interim disposition that preserves the current application stack while deferring the SaaS transition.
Right-Sizing Notes
Why the cloud sizing estimate is set where it is, based on the telemetry available.
Fleet average memory active utilization is 2.5% (penrad-v: 2.0%, penradtest-v: 3.0%). Under Rule T3 with adequate telemetry depth, this reading would qualify for SRC-RS-004 (ZOMBIE). However, Rule T2 fires unconditionally because snapshot_count=1 and apm_telemetry_available=false. A single point-in-time snapshot cannot distinguish a genuinely idle VM from one sampled between workload bursts; this is especially relevant for a radiology imaging application (PenRad) whose workload is inherently episodic and batch-oriented. Maximum allowable tier is therefore SRC-RS-002 (MODERATE, 0.25 reduction). CPU utilization data is entirely absent (cpu_avg_pct null on all 2 VMs), further limiting analytical confidence. Rule T1 does not apply: application_type is Client/Server, not Hardware-Integrated. Both VMs run Windows Server 2019 (EOL January 2029), which is not end-of-life; no HIPAA-EOL flags are warranted. Right-sizing opportunity is flagged at MODERATE tier pending collection of multi-snapshot or APM-backed telemetry to confirm or escalate the tier.
This tab shows the raw output of each analysis agent — the specialized components that examine telemetry, dependencies, contracts, and cost. It is the most detailed and most technical view. Most readers will find the Overview and Backlog tabs sufficient.
Telemetry Agent
Utilization Profilemoderate
Snapshot Count1
APM AvailableFalse
Hard Rule TriggeredTrue
Right-Sizing Opp.True
Agent Confidence45%
Telemetry — VM Counts
Powered Off VMs0
Flagged VMs0
Mem Active % (avg)2.5%
Data WindowNone
Inventory Source
Dependency Agent
Epic IntegrationFalse
PHI HandledTrue
Hardware DependencyFalse
Latency SensitiveFalse
Blast Radiusmedium
Migration RiskMEDIUM
Conf (Inferred)72%
Conf (Confirmed)58%
Verificationsingle_source
Cloud Connectivity
ComplianceHIPAA / HITRUST
Upstream (3)
MUSC-USB Datacenter Network / Storage: confirmed: rvtools.datacenters=['MUSC-USB']
PACS / Radiology Image Archive (inferred): inferred: mammography reporting systems structurally require a PACS or image archive upstream for DICOM image retrieval
HL7 / ADT Feed (inferred): inferred: patient demographic and order data typically fed from HIS/ADT into mammography tracking systems; no catalog…
Downstream (2)
Radiologist Workstations / Reporting Clients: inferred: Client/Server application_type implies end-user client software consuming the PenRad server; mammography…
EMR / Epic (inferred — result delivery): inferred: mammography results are typically routed back to the ordering EMR; catalog.epic_integration=false reduces but…
Procurement Agent
SaaS Available—
SaaS VendorPenRad Technologies
SaaS ProductPenRad Cloud
Repurchase ViableINSUFFICIENT_EVIDENCE
BAA ConfirmedUNKNOWN
SaaS Confidence
Inferred Score70%
Confirmed Score0%
Verificationsingle_source
AWS Equiv.None
Provisioning Agent (Python engine — not LLM)
Current VMs2
Recommended VMs2
Right-Sizing25% reduction
Current TCO/yr$13,000
Projected Cloud TCO$10,400
Migration One-Time$2,080
Dual-Running Cost$7,800
-$7,280
Year-1 Net
$2,600
Steady-State
-$2,080
3-Year
AWS TargetAmazon EC2
Azure TargetAzure Virtual Machines
Cost Modeldefault_multipliers
DEF: {'default_id': 'DEF-WL-001', 'field': 'workload_type', 'value': 'app_server', 'reason': 'workload_type indeterminable'}
DEF: {'default_id': 'DEF-MIG-001', 'field': 'migration_factor', 'value': 0.2, 'reason': 'migration_complexity not provided'}
DEF: {'default_id': 'DEF-COST-001', 'field': 'cost_model_basis', 'value': 'default_multipliers', 'reason': 'no customer cloud quote provided; using SRC-TCO-* and SRC-MULT-* registry defaults'}
Circle
Vendor: Circle CVI · 1 VMs · Criticality: High · HIPAA / HITRUST — PHI handled; BAA requi
REPURCHASE
40%
Confidence
28%
Data Quality
ROI: Year-1 Net: -$3,640 · Steady-State: $1,300/yr · 3-Year: -$1,040
Overview
Backlog
Provenance
Technical Details
Agents
Recommendation
REPURCHASE ⏳ PENDING CONDITIONS
SRC-CAP-001 Binding cap · Ceiling: 40% · Threshold: 60%
Secondary: RETAIN
Confidence
40%
Viable, but specific gaps must close first.
ROI Summary (Python-computed, not LLM)
-$3,640
Year-1 Net
$1,300
Steady-State /yr
-$1,040
3-Year Total
ROI variance: ±60%
⚙ Financials computed by provisioning_engine_v3 (Python) — not LLM-generated
In plain terms: migrating costs money up front — Year 1 is a net outflow of $3,640. From Year 2 onward the app saves $1,300 every year. The 3-Year total nets these against each other.
Finding
Circle CVi Cloud, the vendor's own SaaS offering, is a direct replacement for this on-premises cardiovascular imaging server and eliminates the need to maintain a Windows Server VM at MUSC-USB. However, because this application handles patient imaging data and no Business Associate Agreement has been confirmed for the cloud-hosted product, the migration cannot be authorized until BAA status is resolved and the PACS/RIS integration pathway for the SaaS environment is validated. Once those conditions are met, this is a straightforward vendor-managed SaaS transition.
In plain terms: this app should move to the vendor's cloud product, but a few specific items must be confirmed first. See Conditions to Unlock below.
Conditions to Unlock (3 items)
This recommendation becomes executable once these conditions are confirmed:
☐ Obtain Business Associate Agreement for Circle CVi Cloud SaaS
Owner: Compliance Officer / Legal Counsel · Timeline: 21 days · Unlocks: cap_resolution, compliance_close, migration_commit
☐ Confirm Radiology System Integration Works with Circle CVi Cloud
Owner: Enterprise Architect / Radiology IT Lead · Timeline: 21 days · Unlocks: migration_commit, operational_safety
☐ Install Performance Monitoring on the Circle CVI Server for 30 Days
Owner: Infrastructure / Platform Engineering · Timeline: 35 days · Unlocks: cap_resolution, cost_clarity
See Backlog tab for full action backlog and P2/P3 items.
This is the list of work needed to move this recommendation forward. The 3 P1 items are the priority — they unlock the recommendation. P2 and P3 items improve confidence and reduce risk but are not blockers. 'Story points' measure effort, not time or cost — see the Methodology tab for how they work.
40%
Current
60%
Threshold
55%
Ceiling
34
Story Points
Why This Score
Confidence is hard-capped at 40 by SRC-CAP-001 because phi_handled=true and baa_confirmed=UNKNOWN — the null catalog value for baa_valid cannot be inferred as true or false under v3.1 tri-state rules, and PHI handling without a confirmed BAA is an unconditional ceiling trigger. The pre-cap score of 61.75 reflects a well-supported REPURCHASE signal (vendor SaaS directly named in catalog, hardware_dependency=false, saas_confidence_inferred=70), but that signal cannot be acted upon until the BAA gap is resolved and the three additional caps (SRC-CAP-002, SRC-CAP-003, SRC-CAP-004) are sequentially lifted through the intelligence-gathering backlog.
Proceeding with Circle CVi Cloud migration without a confirmed BAA would constitute a HIPAA violation for a PHI-handling cardiovascular imaging platform; all migration authorization is blocked until APP-029-001 (BAA confirmation) is completed.
Action Backlog — 6 Items · 34 Story Points · Ceiling: 55%
APP-029-001 Confirm BAA Coverage for Circle CVi Cloud (SRC-CAP-001 Resolution)
P1 BAA/Legal
Engage Circle CVI and the MUSC compliance/legal team to determine whether the existing on-premises BAA extends to Circle CVi Cloud or whether a new BAA must be executed for the cloud-hosted product. Obtain a signed BAA document covering PHI processing under the SaaS offering. This is the single binding blocker: SRC-CAP-001 caps confidence at 40 because phi_handled=true and baa_confirmed=UNKNOWN. R
+15% confidence -0% ROI variance 👤 📋 · 8 pts
APP-029-002 Validate PACS/RIS Integration Compatibility with Circle CVi Cloud
P1 Architecture
Engage Circle CVI and the MUSC radiology/IT team to confirm that Circle CVi Cloud supports the existing PACS/RIS integration interfaces (DICOM worklist, DICOM image retrieval, HL7 order context) in a cloud-hosted topology. Document the integration architecture: DICOM connectivity method (DICOMweb vs. traditional DICOM), network path (Direct Connect or VPN), and any PACS/RIS configuration changes r
+0% confidence -0% ROI variance 👤 📋 · 8 pts
APP-029-003 Deploy 30-Day APM Telemetry Collection on circlecviprd
P1 Telemetry
Deploy an APM agent (e.g., Dynatrace, AppDynamics, or equivalent) on the circlecviprd Windows Server 2019 VM to collect CPU and RAM time-series data over a minimum 30-day window. Current data is a single RVTools point-in-time snapshot (snapshot_count=1, apm_telemetry_available=false), which triggers SRC-CAP-004 (ceiling 65) and the RULE T2 hard cap preventing OVERSIZED or ZOMBIE right-sizing class
+0% confidence -0% ROI variance 👤 📋 · 5 pts
APP-029-004 Obtain Circle CVi Cloud Contract Terms and SaaS Pricing
P2 Procurement
Engage Circle CVI procurement to obtain: (1) customer-specific SaaS pricing for Circle CVi Cloud, (2) contract ACV and renewal date for the existing on-premises license, (3) migration path and transition terms from on-premises cvi42 to cloud-hosted equivalent, and (4) data residency and hosting region options (AWS/Azure) for HIPAA compliance. Contract intelligence is currently INSUFFICIENT_EVIDENC
+0% confidence -0% ROI variance 👤 📋 · 5 pts
APP-029-005 Confirm Life-Safety Classification for Circle CVI Cardiovascular Imaging
P2 Security/Compliance
Engage the MUSC clinical informatics and radiology leadership (contact: Wendy Ketchum) to formally confirm or deny life_safety_classification for Circle CVI. Cardiovascular imaging analysis software may carry life-safety implications depending on clinical workflow (e.g., real-time cardiac quantification used in interventional decision-making vs. retrospective reporting). Currently UNKNOWN per v3.1
+0% confidence -0% ROI variance 👤 📋 · 3 pts
APP-029-006 Document DR Requirements and Map to Circle CVi Cloud SLA
P2 Architecture
Work with MUSC IT and radiology operations to document RTO/RPO requirements for this High-criticality PHI-handling cardiovascular imaging system. DR requirements are currently MISSING, applying a 10% penalty to ROI variance. Once RTO/RPO are defined, map them against Circle CVi Cloud SLA terms (to be obtained in APP-029-004) to confirm the SaaS offering meets DR obligations. If Circle CVi Cloud SL
+0% confidence -0% ROI variance 👤 📋 · 5 pts
This tab shows the evidence behind the recommendation. The seven dimensions are scored by quality — CONFIRMED evidence counts fully, INFERRED evidence counts less. The validator table confirms each analysis step passed its checks. Green means the step is sound.
Evidence Classification — 7 Dimensions
Dimension Classification Multiplier Contribution
Telemetry SINGLE 0.50 6.00
Dependency Mapping INFERRED 0.25 2.50
Procurement Contracts MISSING 0.00 0.00
Architecture Diagrams MISSING 0.00 0.00
Security Compliance INFERRED 0.25 1.25
Business Criticality SINGLE 0.50 2.00
Dr Requirements MISSING 0.00 0.00
Applied Cap: SRC-CAP-001
Validation Reports — Per Agent
Agent Status Provenance Counts Defaults Used Rule Failures
telemetry PASS
SRC:2CUST:10COMP:3DEF:2
 None ✓ None
dependency PASS
SRC:9CUST:1COMP:3DEF:6
 None ✓ None
procurement PASS
None ✓ None
provisioning PASS_WITH_DEFAULTS
SRC:7CUST:1COMP:9DEF:1
• {'default_id': 'DEF-WL-001', 'field': 'workload_type', 'value': 'app_server', 'r
• {'default_id': 'DEF-MIG-001', 'field': 'migration_factor', 'value': 0.2, 'reason
 ✓ None
synthesizer PASS
None ✓ None
confidence_advisor PASS
COMP:15
 None ✓ None
This section contains the full technical reasoning behind the recommendation: the rules that fired, the decision path, the confidence math, and the open intelligence gaps. It is intended for analysts and engineers. Executive readers can rely on the Overview tab.
Rules Applied
These are the specific governance rules that produced this recommendation.
▸ RULE 1 — REPURCHASE
A vendor SaaS replacement is available and credible, and the app has no hardware constraint — so the recommendation is to replace it with the vendor's cloud service.
▸ SRC-CAP-001 (binding)
Confidence is capped at 40 because the app handles PHI and the vendor's Business Associate Agreement is not confirmed. PHI cannot move to a cloud vendor without a confirmed BAA.
▸ SRC-CAP-002
Confidence is capped at 55 because more than half of the seven evidence dimensions are inferred or missing — the conclusion is data-thin.
▸ SRC-CAP-003
Confidence is capped at 65 because none of the seven evidence dimensions reached CONFIRMED quality — there is no high-quality anchor for the analysis.
▸ SRC-CAP-004
Confidence is capped at 65 because telemetry is a single point-in-time snapshot with no ongoing monitoring — right-sizing needs usage patterns over time.
▸ RULE T2 — Right-Sizing Hard Cap
Right-sizing is limited to a moderate 25% reduction because telemetry is a single snapshot. More aggressive right-sizing needs 30+ days of monitoring data.
Decision Path
The step-by-step trace of how the recommendation was reached, in the system's own terms.
PRE-CHECK: hardware_dependency=false → structural blocker does not fire. RULE 2 (RETIRE): no decommission annotation, no named successor → does not fire. RULE 1 (REPURCHASE): saas_alternative_available=true (Circle CVi Cloud, catalog notes + procurement), saas_confidence_inferred=70 ≥ 50, hardware_dependency=false → FIRES. strategic_recommendation=REPURCHASE. CONFIDENCE: base=50, telemetry(SINGLE)=+6.0, dependency_mapping(INFERRED)=+2.5, procurement_contracts(MISSING)=+0.0, architecture_diagrams(MISSING)=+0.0, security_compliance(INFERRED)=+1.25, business_criticality(SINGLE)=+2.0, dr_requirements(MISSING)=+0.0 → pre-cap=61.75. CAPS: SRC-CAP-001(phi_handled=true, baa_confirmed=UNKNOWN)→40; SRC-CAP-002(5/7 dims INFERRED/MISSING)→55; SRC-CAP-003(zero CONFIRMED)→65; SRC-CAP-004(snapshot_count=1, apm=false)→65. Lowest cap=40. final_confidence=40. GATE (v4.1): final_confidence=40 < 60, strategic_recommendation=REPURCHASE (not RETAIN) → recommendation_status=PENDING_CONDITIONS, displayed_primary=REPURCHASE (gate does NOT force RETAIN), confidence_gate_applied=true.
Cap Evaluation (all fired caps): SRC-CAP-001: PHI handled + BAA not confirmed → ceiling 40 (ceiling=40) ← BINDING SRC-CAP-002: ≥4 of 7 dimensions INFERRED/MISSING → ceiling 55 (ceiling=55) SRC-CAP-003: Zero CONFIRMED dimensions → ceiling 65 (ceiling=65) SRC-CAP-004: Single snapshot, no APM → ceiling 65 (ceiling=65)
Confidence Scoring Breakdown
How the confidence percentage was calculated, dimension by dimension.
Base score: 50 Evidence contributions: Telemetry SINGLE +6.00 Dependency Mapping INFERRED +2.50 Procurement Contracts MISSING +0.00 Architecture Diagrams MISSING +0.00 Security Compliance INFERRED +1.25 Business Criticality SINGLE +2.00 Dr Requirements MISSING +0.00 Pre-cap score: 61.75 Binding cap applied: SRC-CAP-001 Final confidence: 40%
Intelligence Gaps
Specific things we don't yet know. Each gap is a reason the confidence score is held back.
BAA Confirmation for Circle CVi Cloud
SRC-CAP-001 caps confidence at 40. PHI is handled by this cardiovascular imaging platform; no SaaS migration can be authorized without a signed BAA covering Circle CVi Cloud. This is the binding blocker for REPURCHASE execution.
PACS / RIS Integration Compatibility with Circle CVi Cloud
Cardiovascular imaging workflows depend on DICOM feeds from PACS and order context from RIS. The SaaS migration path requires confirmation that Circle CVi Cloud supports the existing PACS/RIS integration interfaces (DICOM, HL7) in a cloud-hosted topology. Unconfirmed integration breaks the repurchase path.
APM Telemetry Absent
apm_telemetry_available=false with snapshot_count=1 means right-sizing and performance baselining rely solely on a single RVTools snapshot. SRC-CAP-004 applies. Memory active at 7% suggests over-provisioning but cannot be confirmed without longitudinal telemetry.
Contract Intelligence Missing
contract_intelligence=INSUFFICIENT_EVIDENCE. Current on-premises license terms, renewal dates, and SaaS transition pricing are unknown. Procurement penalty of 15% applied to ROI variance. Contract review required before migration commitment.
Architecture Diagrams Absent
No architecture diagrams available. Downstream report delivery mechanism (HL7/FHIR to EMR) is unconfirmed. Architecture penalty of 15% applied to ROI variance. Integration map required for migration planning.
Latency Sensitivity and Life-Safety Classification Unconfirmed
Both latency_sensitive and life_safety_classification are UNKNOWN. Cardiovascular imaging platforms may carry life-safety implications depending on clinical workflow (e.g., real-time cardiac analysis). If either is confirmed true, migration risk escalates and structural constraints must be re-evaluated.
DR Requirements Undocumented
No DR documentation for a High-criticality PHI-handling system. RTO/RPO requirements unknown. DR penalty of 10% applied to ROI variance. DR requirements must be mapped to Circle CVi Cloud SLA before migration.
Technical Detail
A fuller technical explanation of the recommendation and its constraints.
Rule 1 (REPURCHASE) fires: saas_alternative_available=true (Circle CVi Cloud per catalog notes and procurement output), saas_confidence_inferred=70 ≥ 50, and hardware_dependency=false. The pre-check structural blocker does not apply as hardware_dependency=false and neither latency_sensitive nor life_safety_classification is confirmed true (both UNKNOWN). Procurement confirms the SaaS product is the same vendor's cloud-hosted offering, making this a direct lift to vendor-managed SaaS rather than a platform change. Confidence is capped at 40 by SRC-CAP-001 (phi_handled=true, baa_confirmed=UNKNOWN), triggering PENDING_CONDITIONS status; the strategic recommendation remains REPURCHASE and is surfaced to stakeholders with conditions to unlock.
Secondary Path
The fallback option if the primary recommendation cannot proceed.
Secondary Path — RETAIN
RETAIN is the secondary disposition if BAA cannot be confirmed or PACS/RIS integration with Circle CVi Cloud proves technically infeasible, preserving the current stable on-premises footprint.
Right-Sizing Notes
Why the cloud sizing estimate is set where it is, based on the telemetry available.
Single point-in-time snapshot (snapshot_count=1, apm_telemetry_available=false) triggers RULE T2 hard cap. Raw fleet_avg_mem_active_pct=7.0% falls below the 25% OVERSIZED threshold and approaches ZOMBIE territory under RULE T3 thresholds; however, RULE T2 refusal trigger applies — single-snapshot data cannot distinguish a genuinely idle VM from one sampled between bursts. Tier is capped at SRC-RS-002 (MODERATE, 0.25 reduction). SRC-RS-003 (OVERSIZED) and SRC-RS-004 (ZOMBIE) are FORBIDDEN for this run. CPU utilization data is entirely absent (cpu_avg_pct null on 1/1 VMs, no APM). Recommend collection of 30+ days of CPU and RAM time-series telemetry before escalating right-sizing recommendation beyond MODERATE. OS (Windows Server 2019) is within support lifecycle; no EOL compliance flags generated.
This tab shows the raw output of each analysis agent — the specialized components that examine telemetry, dependencies, contracts, and cost. It is the most detailed and most technical view. Most readers will find the Overview and Backlog tabs sufficient.
Telemetry Agent
Utilization Profilemoderate
Snapshot Count1
APM AvailableFalse
Hard Rule TriggeredTrue
Right-Sizing Opp.True
Agent Confidence40%
Telemetry — VM Counts
Powered Off VMs0
Flagged VMs0
Mem Active % (avg)7.0%
Data WindowNone
Inventory Source
Dependency Agent
Epic IntegrationFalse
PHI HandledTrue
Hardware DependencyFalse
Latency SensitiveUNKNOWN
Blast Radiushigh
Migration RiskMEDIUM
Conf (Inferred)72%
Conf (Confirmed)58%
Verificationsingle_source
Cloud Connectivity
ComplianceHIPAA / HITRUST
Upstream (2)
PACS / Radiology Information System (RIS): inferred: cardiovascular imaging platform requires DICOM image feed from PACS and order/patient context from RIS;…
Active Directory / Identity Provider: inferred: Windows Server 2019 Client/Server application in enterprise healthcare environment; standard AD dependency…
Downstream (2)
Cardiology / Radiology Clinical Workflow (reading physicians): inferred: cardiovascular imaging analysis platform; output consumed by cardiologists/radiologists for diagnostic…
EMR / Epic (report integration): inferred: epic_integration=false per catalog, but diagnostic reports from cardiovascular imaging platforms are…
Procurement Agent
SaaS Available—
SaaS VendorCircle CVI
SaaS ProductCircle CVi Cloud
Repurchase ViableINSUFFICIENT_EVIDENCE
BAA ConfirmedUNKNOWN
SaaS Confidence
Inferred Score70%
Confirmed Score0%
Verificationsingle_source
AWS Equiv.None
Provisioning Agent (Python engine — not LLM)
Current VMs1
Recommended VMs1
Right-Sizing25% reduction
Current TCO/yr$6,500
Projected Cloud TCO$5,200
Migration One-Time$1,040
Dual-Running Cost$3,900
-$3,640
Year-1 Net
$1,300
Steady-State
-$1,040
3-Year
AWS TargetAmazon EC2
Azure TargetAzure Virtual Machines
Cost Modeldefault_multipliers
DEF: {'default_id': 'DEF-WL-001', 'field': 'workload_type', 'value': 'app_server', 'reason': 'workload_type indeterminable'}
DEF: {'default_id': 'DEF-MIG-001', 'field': 'migration_factor', 'value': 0.2, 'reason': 'migration_complexity not provided'}
DEF: {'default_id': 'DEF-COST-001', 'field': 'cost_model_basis', 'value': 'default_multipliers', 'reason': 'no customer cloud quote provided; using SRC-TCO-* and SRC-MULT-* registry defaults'}
Nuance PowerShare
Vendor: Nuance (Microsoft) · 10 VMs · Criticality: High · HIPAA / HITRUST
REPURCHASE
40%
Confidence
42%
Data Quality
ROI: Year-1 Net: -$20,453 · Steady-State: $23,400/yr · 3-Year: $26,347
Overview
Backlog
Provenance
Technical Details
Agents
Recommendation
REPURCHASE ⏳ PENDING CONDITIONS
SRC-CAP-001 Binding cap · Ceiling: 40% · Threshold: 60%
Secondary: REHOST
Confidence
40%
Viable, but specific gaps must close first.
ROI Summary (Python-computed, not LLM)
-$20,453
Year-1 Net
$23,400
Steady-State /yr
$26,347
3-Year Total
ROI variance: ±60%
⚙ Financials computed by provisioning_engine_v3 (Python) — not LLM-generated
In plain terms: migrating costs money up front — Year 1 is a net outflow of $20,453. From Year 2 onward the app saves $23,400 every year. The 3-Year total nets these against each other.
Finding
Nuance PowerShare has a direct vendor-hosted SaaS replacement — Nuance PowerShare Cloud — offered by the same vendor (Nuance/Microsoft), which would eliminate all 10 on-premises VMs and reduce operational overhead. However, because this application handles protected health information and no Business Associate Agreement has been confirmed for the cloud service, the migration cannot be executed until BAA status is resolved. Once BAA confirmation and contract terms are in hand, this is a straightforward same-vendor SaaS transition with a clear migration POC already identified in Radiology.
In plain terms: this app should move to the vendor's cloud product, but a few specific items must be confirmed first. See Conditions to Unlock below.
Conditions to Unlock (2 items)
This recommendation becomes executable once these conditions are confirmed:
☐ Verify Business Associate Agreement for Cloud Migration
Owner: Legal / Compliance Officer + Microsoft Account Manager · Timeline: 10 days · Unlocks: cap_resolution, compliance_close, migration_commit
☐ Get Current Contract Terms and Cloud Pricing for PowerShare
Owner: Procurement / Microsoft Account Manager · Timeline: 15 days · Unlocks: cost_clarity, migration_commit
See Backlog tab for full action backlog and P2/P3 items.
This is the list of work needed to move this recommendation forward. The 2 P1 items are the priority — they unlock the recommendation. P2 and P3 items improve confidence and reduce risk but are not blockers. 'Story points' measure effort, not time or cost — see the Methodology tab for how they work.
40%
Current
60%
Threshold
55%
Ceiling
32
Story Points
Why This Score
Confidence is capped at 40 by SRC-CAP-001 (phi_handled=true, baa_confirmed=UNKNOWN) — the single most restrictive cap in the stack — regardless of the pre-cap score of 76.0, because HIPAA prohibits PHI migration to any cloud service without a confirmed BAA. Secondary caps SRC-CAP-002 (ceiling 55, majority INFERRED dimensions) and SRC-CAP-004 (ceiling 65, single snapshot, no APM) are also active but subordinate to SRC-CAP-001 as the binding constraint.
Proceeding with REPURCHASE to Nuance PowerShare Cloud without a confirmed BAA exposes the organization to HIPAA enforcement risk and potential OCR investigation for unauthorized PHI disclosure to a Business Associate without a compliant agreement.
Action Backlog — 6 Items · 32 Story Points · Ceiling: 55%
APP-030-001 Confirm BAA Status for Nuance PowerShare Cloud (Microsoft/Nuance Entity)
P1 BAA/Legal
Obtain and review the executed Business Associate Agreement (BAA) covering Nuance PowerShare Cloud under the post-acquisition Microsoft entity. The Nuance-to-Microsoft acquisition (2022) requires verification that any legacy BAA with Nuance Communications, Inc. has been novated or replaced by a Microsoft Corporation BAA. Confirm the BAA explicitly covers Nuance PowerShare Cloud (vendor-hosted SaaS
+15% confidence -0% ROI variance 👤 📋 · 5 pts
APP-030-002 Obtain Current Licensing Contract and SaaS Migration Pricing from Nuance/Microsoft
P1 Procurement
Engage the Nuance/Microsoft account team to obtain: (1) the current on-premises PowerShare license agreement including ACV, renewal date, and termination/migration clauses; (2) a formal quote or order form for Nuance PowerShare Cloud including per-study or subscription pricing; (3) any enterprise agreement migration credits or co-term options available under the Microsoft EA. This resolves contrac
+0% confidence -0% ROI variance 👤 📋 · 5 pts
APP-030-003 Deploy 30-Day APM Telemetry Collection Across All 10 PowerShare VMs
P2 Telemetry
Install and configure APM or performance monitoring agents (e.g., Azure Monitor, Datadog, SolarWinds) on all 10 Nuance PowerShare VMs to collect 30+ days of CPU and RAM time-series data. Priority targets are pshareflorence (120 GiB RAM, 1.0% mem_active at snapshot) and npsaccgwprd (64 GiB RAM, 1.0% mem_active at snapshot), which together account for 184 GiB of the 328 GiB fleet total. A single RVT
+0% confidence -0% ROI variance 👤 📋 · 8 pts
APP-030-004 Produce Architecture Diagram — PACS Integration, Gateway Topology, and Florence Site Configuration
P2 Architecture
Engage the Radiology IT team (POC: Eric Schneider, Michael Slezak per VM notes) to produce a confirmed architecture diagram covering: (1) PACS/modality feed topology into PowerShare (npsaccgwprd/npsaccgwtst gateway layer); (2) Florence site connectivity and pshareflorence VM role — dedicated site instance vs. enterprise instance routing; (3) external referring provider access patterns and network
+0% confidence -0% ROI variance 👤 📋 · 8 pts
APP-030-005 Document DR/RTO-RPO Requirements and Validate Against Nuance PowerShare Cloud SLA
P3 Architecture
Obtain the formal DR/RTO-RPO requirements for Nuance PowerShare from the clinical operations or IT governance team. High criticality classification implies documented RTO/RPO requirements exist but are currently unconfirmed (dr_requirements=INFERRED). Compare confirmed requirements against the Nuance PowerShare Cloud SLA (uptime, RPO, RTO commitments). Verify that the SaaS SLA meets or exceeds cli
+0% confidence -0% ROI variance 👤 📋 · 3 pts
APP-030-006 Develop Florence Site Cutover Plan and Validate Connectivity for SaaS Migration
P3 Dependency Mapping
The pshareflorence VM (120 GiB RAM, Windows Server 2016, 1.0% mem_active) serves the Florence site as a dedicated PowerShare instance (requestor: Wendy Ketchum per VM notes). A SaaS migration must account for: (1) Florence site network connectivity and bandwidth adequacy for cloud-hosted image sharing; (2) any site-specific configuration or local caching requirements currently served by the dedica
+0% confidence -0% ROI variance 👤 📋 · 3 pts
This tab shows the evidence behind the recommendation. The seven dimensions are scored by quality — CONFIRMED evidence counts fully, INFERRED evidence counts less. The validator table confirms each analysis step passed its checks. Green means the step is sound.
Evidence Classification — 7 Dimensions
Dimension Classification Multiplier Contribution
Telemetry SINGLE 0.50 6.00
Dependency Mapping CONFIRMED 1.00 10.00
Procurement Contracts INFERRED 0.25 2.00
Architecture Diagrams INFERRED 0.25 1.75
Security Compliance INFERRED 0.25 1.25
Business Criticality CONFIRMED 1.00 4.00
Dr Requirements INFERRED 0.25 1.00
Applied Cap: SRC-CAP-001
Validation Reports — Per Agent
Agent Status Provenance Counts Defaults Used Rule Failures
telemetry PASS
SRC:2CUST:9COMP:4DEF:2
 None ✓ None
dependency PASS
SRC:9COMP:5DEF:3
 None ✓ None
procurement PASS
None ✓ None
provisioning PASS_WITH_DEFAULTS
SRC:7CUST:1COMP:9DEF:1
• {'default_id': 'DEF-WL-001', 'field': 'workload_type', 'value': 'app_server', 'r
• {'default_id': 'DEF-MIG-001', 'field': 'migration_factor', 'value': 0.2, 'reason
 ✓ None
synthesizer PASS
None ✓ None
confidence_advisor PASS
COMP:33
 None ✓ None
This section contains the full technical reasoning behind the recommendation: the rules that fired, the decision path, the confidence math, and the open intelligence gaps. It is intended for analysts and engineers. Executive readers can rely on the Overview tab.
Rules Applied
These are the specific governance rules that produced this recommendation.
▸ RULE 1 — REPURCHASE
A vendor SaaS replacement is available and credible, and the app has no hardware constraint — so the recommendation is to replace it with the vendor's cloud service.
▸ SRC-CAP-001 (binding)
Confidence is capped at 40 because the app handles PHI and the vendor's Business Associate Agreement is not confirmed. PHI cannot move to a cloud vendor without a confirmed BAA.
▸ SRC-CAP-002
Confidence is capped at 55 because more than half of the seven evidence dimensions are inferred or missing — the conclusion is data-thin.
▸ SRC-CAP-004
Confidence is capped at 65 because telemetry is a single point-in-time snapshot with no ongoing monitoring — right-sizing needs usage patterns over time.
▸ RULE T2 — Right-Sizing Hard Cap
Right-sizing is limited to a moderate 25% reduction because telemetry is a single snapshot. More aggressive right-sizing needs 30+ days of monitoring data.
Decision Path
The step-by-step trace of how the recommendation was reached, in the system's own terms.
PRE-CHECK: hardware_dependency=false, life_safety_classification=false → does not fire. RULE 2 (RETIRE): no decommission_annotation_present, no named successor → does not fire. RULE 1 (REPURCHASE): saas_alternative_available=true (Nuance PowerShare Cloud per catalog notes), saas_confidence_inferred=80 >= 50, hardware_dependency=false → FIRES. strategic_recommendation=REPURCHASE. CONFIDENCE SCORING: base=50 + telemetry(SINGLE,12×0.50=6.0) + dependency(CONFIRMED,10×1.00=10.0) + procurement(INFERRED,8×0.25=2.0) + architecture(INFERRED,7×0.25=1.75) + security(INFERRED,5×0.25=1.25) + criticality(CONFIRMED,4×1.00=4.0) + dr(INFERRED,4×0.25=1.0) = 76.0 pre-cap. CAP EVALUATION: SRC-CAP-001 fires (phi_handled=true, baa_confirmed=UNKNOWN) → ceiling 40; SRC-CAP-002 fires (4/7 INFERRED = majority) → ceiling 55; SRC-CAP-004 fires (snapshot_count=1, apm=false) → ceiling 65. Lowest cap = SRC-CAP-001 → final_confidence=40. GATE (v4.1): final_confidence=40 < 60, strategic_recommendation=REPURCHASE (not RETAIN) → recommendation_status=PENDING_CONDITIONS, displayed_primary=REPURCHASE, confidence_gate_applied=true.
Cap Evaluation (all fired caps): SRC-CAP-001: PHI handled + BAA not confirmed → ceiling 40 (ceiling=40) ← BINDING SRC-CAP-002: ≥4 of 7 dimensions INFERRED/MISSING → ceiling 55 (ceiling=55) SRC-CAP-004: Single snapshot, no APM → ceiling 65 (ceiling=65)
Confidence Scoring Breakdown
How the confidence percentage was calculated, dimension by dimension.
Base score: 50 Evidence contributions: Telemetry SINGLE +6.00 Dependency Mapping CONFIRMED +10.00 Procurement Contracts INFERRED +2.00 Architecture Diagrams INFERRED +1.75 Security Compliance INFERRED +1.25 Business Criticality CONFIRMED +4.00 Dr Requirements INFERRED +1.00 Pre-cap score: 76.0 Binding cap applied: SRC-CAP-001 Final confidence: 40%
Intelligence Gaps
Specific things we don't yet know. Each gap is a reason the confidence score is held back.
BAA Confirmation for Nuance PowerShare Cloud
SRC-CAP-001 binding cap (ceiling 40) cannot be lifted until BAA is confirmed for the SaaS target. PHI is handled; HIPAA requires a signed BAA with Nuance/Microsoft before any cloud transition. This is the single highest-priority unlock condition.
Contract Intelligence — Current Licensing and SaaS Transition Terms
contract_intelligence=INSUFFICIENT_EVIDENCE. Current on-premises licensing terms, renewal dates, and SaaS migration pricing are unknown. Without this, ROI projection carries maximum variance (60%). Procurement must obtain current contract and SaaS quote.
APM / Application Performance Telemetry
apm_telemetry_available=false; single RVTools snapshot only. Memory active utilization is very low (avg 5.8%) but without APM data, transaction volumes, peak load patterns, and SLA requirements for image sharing cannot be validated. Required for right-sizing the SaaS tier.
Architecture Diagram — PACS Integration and Gateway Configuration
No architecture diagrams available. The gateway/interface layer (npsaccgwprd, npsaccgwtst) and PACS feed topology are inferred from VM naming. SaaS migration requires confirmed integration architecture to assess cutover complexity for Florence site and external referring providers.
DR / RTO-RPO Requirements
No DR documentation cited. High criticality classification implies RTO/RPO requirements exist but are unconfirmed. SaaS SLA must be validated against clinical DR requirements before transition.
Florence Site Dependency Cutover Plan
pshareflorence VM indicates a dedicated site dependency. SaaS migration must account for Florence site connectivity, latency, and any site-specific configuration before decommissioning the dedicated VM.
Technical Detail
A fuller technical explanation of the recommendation and its constraints.
Rule 1 (REPURCHASE) fires: saas_alternative_available=true (Nuance PowerShare Cloud, catalog notes), saas_confidence_inferred=80 >= 50, hardware_dependency=false. The dependency agent confirms no hardware or life-safety constraints, and the VM notes (PshareENTPRD: 'pShareFlorence replacement POC') indicate active internal evaluation of the SaaS path. SRC-CAP-001 applies as the binding cap (phi_handled=true, baa_confirmed=UNKNOWN), reducing final_confidence to 40 and triggering PENDING_CONDITIONS status. The strategic recommendation remains REPURCHASE per v4.1 governance — the gate surfaces conditions rather than overriding disposition.
Secondary Path
The fallback option if the primary recommendation cannot proceed.
Secondary Path — REHOST
REHOST is the secondary disposition if BAA confirmation cannot be obtained in the near term; the 10 VMs have no hardware dependency and are standard lift-and-shift candidates, though the mixed OS estate (2016/2022/2025) would require OS remediation prior to migration.
Right-Sizing Notes
Why the cloud sizing estimate is set where it is, based on the telemetry available.
Fleet avg mem_active_pct = 5.8% (CUSTOMER_PROVIDED, RVTools point-in-time). Raw utilization would qualify for SRC-RS-003 OVERSIZED under RULE T3 (5.8% < 25%). However, snapshot_count=1 AND apm_telemetry_available=false unconditionally triggers RULE T2: OVERSIZED and ZOMBIE tiers are FORBIDDEN; tier is capped at SRC-RS-002 MODERATE (0.25 reduction). A single snapshot cannot distinguish genuinely idle VMs from those sampled between burst cycles. Two VMs warrant attention if additional telemetry is collected: pshareflorence (120 GiB RAM, 1.0% mem_active_pct — PshareENTPRD notes identify this as a Florence replacement POC target) and npsaccgwprd (64 GiB RAM, 1.0% mem_active_pct). These two VMs account for 184 GiB of the 328 GiB fleet total. No EOL OS violations detected; all OS versions are within vendor support windows. phi_handling=true noted; no HIPAA-EOL flags triggered.
This tab shows the raw output of each analysis agent — the specialized components that examine telemetry, dependencies, contracts, and cost. It is the most detailed and most technical view. Most readers will find the Overview and Backlog tabs sufficient.
Telemetry Agent
Utilization Profilemoderate
Snapshot Count1
APM AvailableFalse
Hard Rule TriggeredTrue
Right-Sizing Opp.True
Agent Confidence45%
Telemetry — VM Counts
Powered Off VMs0
Flagged VMs0
Mem Active % (avg)5.8%
Data WindowNone
Inventory Source
Dependency Agent
Epic IntegrationFalse
PHI HandledTrue
Hardware DependencyFalse
Latency SensitiveFalse
Blast Radiushigh
Migration RiskMEDIUM
Conf (Inferred)78%
Conf (Confirmed)72%
Verificationconfirmed
Cloud Connectivity
ComplianceHIPAA / HITRUST
Upstream (3)
Radiology PACS / Modality Systems: inferred: vm_notes reference 'POWERSHARE server for Radiology' and 'pShareFlorence replacement POC: Eric Schneider…
SQL Database Backend (npssqlprd / npssqltst): confirmed: rvtools.vm_notes — 'Nuance PowerShare SQL' and 'Nuance PowerShare Test SQL'
Network Gateway / Interface Layer (npsaccgwprd / npsaccgwtst): confirmed: rvtools.vm_notes — 'Nuance PowerShare Gateway/Interface' (both prod and test)
Downstream (3)
Radiology Department Consumers (Radiologists, Referring Physicians): inferred: PowerShare is an image distribution platform; downstream consumers are clinical staff accessing shared…
Florence Site (pshareflorence): confirmed: rvtools.vm_notes — 'Powershare for Florence Requestor: Wendy Ketchum'; dedicated VM indicates a downstream…
External Referring Providers / Health System Partners: inferred: PowerShare's core function is cross-organizational image sharing; external provider access is a structural…
Procurement Agent
SaaS Available—
SaaS VendorNuance (Microsoft)
SaaS ProductNuance PowerShare Cloud
Repurchase ViableINSUFFICIENT_EVIDENCE
BAA ConfirmedUNKNOWN
SaaS Confidence
Inferred Score80%
Confirmed Score0%
Verificationsingle_source
AWS Equiv.None
Provisioning Agent (Python engine — not LLM)
Current VMs10
Recommended VMs8
Right-Sizing25% reduction
Current TCO/yr$65,000
Projected Cloud TCO$41,600
Migration One-Time$8,320
Dual-Running Cost$35,533
-$20,453
Year-1 Net
$23,400
Steady-State
$26,347
3-Year
AWS TargetAmazon EC2
Azure TargetAzure Virtual Machines
Cost Modeldefault_multipliers
DEF: {'default_id': 'DEF-WL-001', 'field': 'workload_type', 'value': 'app_server', 'reason': 'workload_type indeterminable'}
DEF: {'default_id': 'DEF-MIG-001', 'field': 'migration_factor', 'value': 0.2, 'reason': 'migration_complexity not provided'}
DEF: {'default_id': 'DEF-COST-001', 'field': 'cost_model_basis', 'value': 'default_multipliers', 'reason': 'no customer cloud quote provided; using SRC-TCO-* and SRC-MULT-* registry defaults'}
Nuance PowerScribe One
Vendor: Nuance (Microsoft) · 1 VMs · Criticality: High · HIPAA / HITRUST
REPURCHASE
40%
Confidence
28%
Data Quality
ROI: Year-1 Net: -$3,640 · Steady-State: $1,300/yr · 3-Year: -$1,040
Overview
Backlog
Provenance
Technical Details
Agents
Recommendation
REPURCHASE ⏳ PENDING CONDITIONS
SRC-CAP-001 Binding cap · Ceiling: 40% · Threshold: 60%
Secondary: REHOST
Confidence
40%
Viable, but specific gaps must close first.
ROI Summary (Python-computed, not LLM)
-$3,640
Year-1 Net
$1,300
Steady-State /yr
-$1,040
3-Year Total
ROI variance: ±50%
⚙ Financials computed by provisioning_engine_v3 (Python) — not LLM-generated
In plain terms: migrating costs money up front — Year 1 is a net outflow of $3,640. From Year 2 onward the app saves $1,300 every year. The 3-Year total nets these against each other.
Finding
Nuance PowerScribe One has a direct same-vendor cloud replacement — Nuance PowerScribe One Cloud — which eliminates the on-premises Windows Server footprint with no hardware dependency blocking migration. However, because this application handles protected health information and no Business Associate Agreement has been confirmed for the cloud product, the REPURCHASE path cannot be executed until BAA status is resolved. Once BAA confirmation and contract terms are in hand, this is a straightforward vendor-managed SaaS transition.
In plain terms: this app should move to the vendor's cloud product, but a few specific items must be confirmed first. See Conditions to Unlock below.
Conditions to Unlock (2 items)
This recommendation becomes executable once these conditions are confirmed:
☐ Confirm Business Associate Agreement for Nuance Cloud Product
Owner: Compliance Officer / Legal Counsel · Timeline: 30 days · Unlocks: compliance_close, cap_resolution, migration_commit
☐ Pull Existing Nuance License Contract and Renewal Terms
Owner: Procurement / Contract Manager · Timeline: 21 days · Unlocks: cost_clarity, migration_commit
See Backlog tab for full action backlog and P2/P3 items.
This is the list of work needed to move this recommendation forward. The 2 P1 items are the priority — they unlock the recommendation. P2 and P3 items improve confidence and reduce risk but are not blockers. 'Story points' measure effort, not time or cost — see the Methodology tab for how they work.
40%
Current
60%
Threshold
55%
Ceiling
26
Story Points
Why This Score
Confidence is hard-capped at 40 by SRC-CAP-001 because phi_handled=true and baa_confirmed=UNKNOWN — a null catalog field that cannot be inferred as true or false under v3.1 tri-state rules. The pre-cap score of 63.5 reflects a well-understood application with a clear SaaS migration path, but no amount of additional evidence can raise confidence above 40 until a signed BAA is confirmed with Nuance (Microsoft) for the cloud product, making BAA confirmation the unconditional first action.
Proceeding with REPURCHASE planning without BAA confirmation exposes the organization to HIPAA enforcement risk for PHI processed under an unconfirmed Business Associate Agreement with the cloud vendor.
Action Backlog — 6 Items · 26 Story Points · Ceiling: 55%
APP-031-001 Confirm BAA Coverage — Nuance PowerScribe One Cloud (Microsoft Entity)
P1 BAA/Legal
Obtain and review a signed Business Associate Agreement from Nuance (Microsoft) explicitly covering PowerScribe One Cloud under the post-2022 Microsoft corporate entity. Confirm the BAA names the organization as a covered entity and that the specific SaaS product line is in scope. This is the single action that resolves SRC-CAP-001 — the current binding confidence cap at 40. Until baa_confirmed tr
+15% confidence -0% ROI variance 👤 📋 · 5 pts
APP-031-002 Retrieve Current On-Premises License Contract — ACV, Renewal Date, SaaS Migration Entitlement
P1 Procurement
Retrieve the current on-premises Nuance PowerScribe One license agreement from the contract management system. Extract: (1) Annual Contract Value (ACV), (2) renewal date, (3) any SaaS migration entitlement or upgrade path language, (4) data portability and termination clauses. This resolves the procurement_contracts=MISSING evidence classification, removes the 15% procurement ROI variance penalty,
+0% confidence -0% ROI variance 👤 📋 · 5 pts
APP-031-003 Map PACS/RIS Integration Topology — HL7/DICOM Feed Confirmation
P2 Dependency Mapping
Confirm and document the upstream PACS/RIS integration topology for Nuance PowerScribe One. Identify: (1) specific PACS and RIS systems feeding HL7 orders and DICOM worklist to PowerScribe One, (2) integration engine or middleware in the path (e.g., Mirth Connect, Rhapsody, Ensemble), (3) network path and port requirements, (4) whether SQL Server on ps1tstsql is co-located on the same VM or a sepa
+0% confidence -0% ROI variance 👤 📋 · 5 pts
APP-031-004 Identify EHR Results Distribution Target — Non-Epic HL7 Routing Endpoint
P2 Dependency Mapping
Identify the downstream EHR or results-routing engine that receives finalized radiology reports from PowerScribe One via HL7. epic_integration=false is confirmed, but the non-Epic EHR target (e.g., Meditech, Cerner, Allscripts, or a results router such as Rhapsody/Mirth) is unidentified. This information is required to re-point result feeds during SaaS cutover and to validate that the SaaS product
+0% confidence -0% ROI variance 👤 📋 · 3 pts
APP-031-005 Deploy 30-Day APM Telemetry Baseline — PowerScribe One Application Layer
P2 Telemetry
Deploy an APM agent (e.g., Dynatrace, AppDynamics, or New Relic) on the PowerScribe One VM (ps1tstsql) to capture a 30-day application-layer performance baseline. Collect: CPU utilization time-series (currently null), memory active percentage across peak radiology workflow periods, response time for dictation and report finalization transactions, and SQL Server query performance. This resolves SRC
+0% confidence -0% ROI variance 👤 📋 · 5 pts
APP-031-006 Document DR Requirements — RTO/RPO for Radiology Reporting Workflow
P3 Architecture
Retrieve or elicit formal RTO and RPO requirements for Nuance PowerScribe One from the Business Continuity / DR program. High criticality designation implies recovery requirements exist but are undocumented (dr_requirements=MISSING). This information is required to: (1) compare on-premises DR posture against Nuance PowerScribe One Cloud SLA commitments, (2) remove the 10% DR ROI variance penalty,
+0% confidence -0% ROI variance 👤 📋 · 3 pts
This tab shows the evidence behind the recommendation. The seven dimensions are scored by quality — CONFIRMED evidence counts fully, INFERRED evidence counts less. The validator table confirms each analysis step passed its checks. Green means the step is sound.
Evidence Classification — 7 Dimensions
Dimension Classification Multiplier Contribution
Telemetry SINGLE 0.50 6.00
Dependency Mapping INFERRED 0.25 2.50
Procurement Contracts MISSING 0.00 0.00
Architecture Diagrams INFERRED 0.25 1.75
Security Compliance INFERRED 0.25 1.25
Business Criticality SINGLE 0.50 2.00
Dr Requirements MISSING 0.00 0.00
Applied Cap: SRC-CAP-001
Validation Reports — Per Agent
Agent Status Provenance Counts Defaults Used Rule Failures
telemetry PASS
SRC:2CUST:9COMP:4DEF:2
 None ✓ None
dependency PASS
CUST:4COMP:5DEF:3
 None ✓ None
procurement PASS
None ✓ None
provisioning PASS_WITH_DEFAULTS
SRC:7CUST:1COMP:9DEF:1
• {'default_id': 'DEF-WL-001', 'field': 'workload_type', 'value': 'app_server', 'r
• {'default_id': 'DEF-MIG-001', 'field': 'migration_factor', 'value': 0.2, 'reason
 ✓ None
synthesizer PASS
None ✓ None
confidence_advisor PASS
COMP:33
 None ✓ None
This section contains the full technical reasoning behind the recommendation: the rules that fired, the decision path, the confidence math, and the open intelligence gaps. It is intended for analysts and engineers. Executive readers can rely on the Overview tab.
Rules Applied
These are the specific governance rules that produced this recommendation.
▸ RULE 1 — REPURCHASE
A vendor SaaS replacement is available and credible, and the app has no hardware constraint — so the recommendation is to replace it with the vendor's cloud service.
▸ SRC-CAP-001 (binding)
Confidence is capped at 40 because the app handles PHI and the vendor's Business Associate Agreement is not confirmed. PHI cannot move to a cloud vendor without a confirmed BAA.
▸ SRC-CAP-002
Confidence is capped at 55 because more than half of the seven evidence dimensions are inferred or missing — the conclusion is data-thin.
▸ SRC-CAP-003
Confidence is capped at 65 because none of the seven evidence dimensions reached CONFIRMED quality — there is no high-quality anchor for the analysis.
▸ SRC-CAP-004
Confidence is capped at 65 because telemetry is a single point-in-time snapshot with no ongoing monitoring — right-sizing needs usage patterns over time.
▸ RULE T2 — Right-Sizing Hard Cap
Right-sizing is limited to a moderate 25% reduction because telemetry is a single snapshot. More aggressive right-sizing needs 30+ days of monitoring data.
Decision Path
The step-by-step trace of how the recommendation was reached, in the system's own terms.
PRE-CHECK: hardware_dependency=false → does not fire. RULE 1 (REPURCHASE): saas_alternative_available=true (catalog notes + procurement), saas_confidence_inferred=80 >= 50, hardware_dependency=false → RULE 1 FIRES. Strategic recommendation = REPURCHASE. CONFIDENCE SCORING: base=50, telemetry=SINGLE(+6.0), dependency_mapping=INFERRED(+2.5), procurement_contracts=MISSING(+0.0), architecture_diagrams=INFERRED(+1.75), security_compliance=INFERRED(+1.25), business_criticality=SINGLE(+2.0), dr_requirements=MISSING(+0.0) → pre-cap=63.5. CAP EVALUATION: SRC-CAP-001 fires (phi_handled=true, baa_confirmed=UNKNOWN) → cap=40; SRC-CAP-002 fires (5/7 dimensions INFERRED/MISSING) → cap=55; SRC-CAP-003 fires (zero CONFIRMED) → cap=65; SRC-CAP-004 fires (snapshot_count=1, apm_telemetry_available=false) → cap=65. Lowest cap = SRC-CAP-001 at 40. Final confidence = 40. GATE (v4.1): final_confidence=40 < 60, strategic_recommendation=REPURCHASE (not RETAIN) → recommendation_status=PENDING_CONDITIONS, displayed_primary=REPURCHASE (gate does NOT force RETAIN in v4.1), confidence_gate_applied=true. ROI VARIANCE: base=5 + telemetry_penalty=0(SINGLE) + procurement_penalty=15(MISSING) + arch_penalty=10(INFERRED) + dep_penalty=10(INFERRED) + dr_penalty=10(MISSING) = 50% (provisioning agent hardcoded 60% assuming all MISSING; corrected to 50% per actual classifications).
Cap Evaluation (all fired caps): SRC-CAP-001: PHI handled + BAA not confirmed → ceiling 40 (ceiling=40) ← BINDING SRC-CAP-002: ≥4 of 7 dimensions INFERRED/MISSING → ceiling 55 (ceiling=55) SRC-CAP-003: Zero CONFIRMED dimensions → ceiling 65 (ceiling=65) SRC-CAP-004: Single snapshot, no APM → ceiling 65 (ceiling=65)
Confidence Scoring Breakdown
How the confidence percentage was calculated, dimension by dimension.
Base score: 50 Evidence contributions: Telemetry SINGLE +6.00 Dependency Mapping INFERRED +2.50 Procurement Contracts MISSING +0.00 Architecture Diagrams INFERRED +1.75 Security Compliance INFERRED +1.25 Business Criticality SINGLE +2.00 Dr Requirements MISSING +0.00 Pre-cap score: 63.5 Binding cap applied: SRC-CAP-001 Final confidence: 40%
Intelligence Gaps
Specific things we don't yet know. Each gap is a reason the confidence score is held back.
BAA Confirmation — Nuance PowerScribe One Cloud
SRC-CAP-001 binding cap at 40. REPURCHASE cannot achieve EXECUTABLE status until a valid BAA is confirmed with Nuance (Microsoft) for the cloud-hosted SaaS product. This is the single highest-priority unlock condition.
Contract Intelligence — Existing On-Premises License
contract_intelligence=INSUFFICIENT_EVIDENCE. Current license terms, renewal date, and SaaS migration entitlement are unknown. Procurement penalty (+15%) applied to ROI variance. Required to validate cost model and migration timeline.
PACS/RIS Integration Mapping
Upstream dependency on Radiology PACS/RIS is inferred from product class, not confirmed. HL7/DICOM feed topology must be documented before SaaS cutover to avoid radiology workflow disruption. Contributes to high blast_radius_estimate.
EHR Results Distribution Topology
Downstream HL7 result routing to EHR/results engine is inferred (epic_integration=false confirmed, but non-Epic EHR target unidentified). SaaS migration requires re-pointing result feeds; target system must be confirmed.
APM / Application Performance Telemetry
apm_telemetry_available=false. Single RVTools snapshot provides memory utilization (9% active) but no application-layer performance data. SRC-CAP-004 cap applies. Right-sizing recommendation (SRC-RS-002, 25% reduction) is directionally valid but unconfirmed without APM baseline.
DR / RTO-RPO Requirements
No DR documentation present. High criticality designation implies recovery requirements exist but are unconfirmed. DR penalty (+10%) applied to ROI variance. Required for SaaS SLA comparison.
SQL Server Co-location Confirmation
SQL Server dependency inferred from VM name suffix 'sql'. Whether this is co-located on the same VM or a separate instance affects migration scope and SaaS data migration planning.
Technical Detail
A fuller technical explanation of the recommendation and its constraints.
The pre-check does not fire: hardware_dependency=false, latency_sensitive=false, life_safety_classification=false. Rule 1 (REPURCHASE) fires because saas_alternative_available=true (confirmed by both catalog notes and procurement output citing Nuance PowerScribe One Cloud), saas_confidence_inferred=80 exceeds the 50 threshold, and hardware_dependency is false. The dependency agent confirms no hardware or life-safety constraints, though it flags high blast radius and migration risk due to inferred PACS/RIS upstream feeds and EHR downstream distribution. The procurement agent confirms the SaaS product exists but cannot confirm BAA or contract terms, which triggers SRC-CAP-001 (phi_handled=true, baa_confirmed=UNKNOWN) capping confidence at 40 — below the 60 threshold, yielding PENDING_CONDITIONS status.
Secondary Path
The fallback option if the primary recommendation cannot proceed.
Secondary Path — REHOST
REHOST to AWS/Azure/GCP IaaS is the fallback if SaaS contract negotiations fail or BAA cannot be executed with Nuance's cloud offering. The single VM (4 vCPU / 24 GiB RAM, 9% memory active) is a straightforward lift-and-shift candidate with a 25% right-sizing opportunity per SRC-RS-002.
Right-Sizing Notes
Why the cloud sizing estimate is set where it is, based on the telemetry available.
RULE T1 does not apply: application_type is Client/Server, not Hardware-Integrated. RULE T2 fires: snapshot_count=1 AND apm_telemetry_available=false. Hard cap enforced at SRC-RS-002 (MODERATE, 0.25 reduction). Raw fleet_avg_mem_active_pct of 9.0% would indicate OVERSIZED or ZOMBIE under RULE T3, but single-snapshot data cannot distinguish a genuinely idle VM from one sampled between burst cycles. ZOMBIE (SRC-RS-004) and OVERSIZED (SRC-RS-003) are FORBIDDEN. CPU utilization data is entirely absent (cpu_avg_pct null on 1 of 1 VMs). Right-sizing opportunity flagged at MODERATE tier pending multi-snapshot or 30-day APM validation. TELEMETRY-DEPTH-INSUFFICIENT flag raised per RULE T2 refusal trigger.
This tab shows the raw output of each analysis agent — the specialized components that examine telemetry, dependencies, contracts, and cost. It is the most detailed and most technical view. Most readers will find the Overview and Backlog tabs sufficient.
Telemetry Agent
Utilization Profilemoderate
Snapshot Count1
APM AvailableFalse
Hard Rule TriggeredTrue
Right-Sizing Opp.True
Agent Confidence45%
Telemetry — VM Counts
Powered Off VMs0
Flagged VMs0
Mem Active % (avg)9.0%
Data WindowNone
Inventory Source
Dependency Agent
Epic IntegrationFalse
PHI HandledTrue
Hardware DependencyFalse
Latency SensitiveFalse
Blast Radiushigh
Migration RiskMEDIUM
Conf (Inferred)72%
Conf (Confirmed)55%
Verificationsingle_source
Cloud Connectivity
ComplianceHIPAA / HITRUST
Upstream (3)
SQL Server (local): inferred: vm_name 'ps1tstsql' contains 'sql' suffix, consistent with co-located or dedicated SQL Server instance for…
Radiology PACS / RIS: inferred: product_class — PowerScribe One is a radiology reporting platform; upstream HL7 or DICOM feeds from PACS/RIS…
Active Directory / Windows Authentication: inferred: OS = Windows Server 2019; Client/Server application type on Windows infrastructure implies AD-integrated…
Downstream (2)
Radiologist / Clinician Workstations: inferred: product_class — PowerScribe One delivers finalized radiology reports to ordering clinicians; downstream…
EHR / Results Distribution (non-Epic): inferred: epic_integration=false (catalog-confirmed); however HL7 result distribution to an EHR or results-routing…
Procurement Agent
SaaS Available—
SaaS VendorNuance (Microsoft)
SaaS ProductNuance PowerScribe One Cloud
Repurchase ViableINSUFFICIENT_EVIDENCE
BAA ConfirmedUNKNOWN
SaaS Confidence
Inferred Score80%
Confirmed Score0%
Verificationsingle_source
AWS Equiv.None
Provisioning Agent (Python engine — not LLM)
Current VMs1
Recommended VMs1
Right-Sizing25% reduction
Current TCO/yr$6,500
Projected Cloud TCO$5,200
Migration One-Time$1,040
Dual-Running Cost$3,900
-$3,640
Year-1 Net
$1,300
Steady-State
-$1,040
3-Year
AWS TargetAmazon EC2
Azure TargetAzure Virtual Machines
Cost Modeldefault_multipliers
DEF: {'default_id': 'DEF-WL-001', 'field': 'workload_type', 'value': 'app_server', 'reason': 'workload_type indeterminable'}
DEF: {'default_id': 'DEF-MIG-001', 'field': 'migration_factor', 'value': 0.2, 'reason': 'migration_complexity not provided'}
DEF: {'default_id': 'DEF-COST-001', 'field': 'cost_model_basis', 'value': 'default_multipliers', 'reason': 'no customer cloud quote provided; using SRC-TCO-* and SRC-MULT-* registry defaults'}
Optellum / Lung AI
Vendor: Optellum · 2 VMs · Criticality: High · HIPAA / HITRUST
REPURCHASE
55%
Confidence
28%
Data Quality
ROI: Year-1 Net: -$7,280 · Steady-State: $2,600/yr · 3-Year: -$2,080
Overview
Backlog
Provenance
Technical Details
Agents
Recommendation
REPURCHASE ⏳ PENDING CONDITIONS
SRC-CAP-002 Binding cap · Ceiling: 55% · Threshold: 60%
Secondary: REHOST
Confidence
55%
Viable, but specific gaps must close first.
ROI Summary (Python-computed, not LLM)
-$7,280
Year-1 Net
$2,600
Steady-State /yr
-$2,080
3-Year Total
ROI variance: ±60%
⚙ Financials computed by provisioning_engine_v3 (Python) — not LLM-generated
In plain terms: migrating costs money up front — Year 1 is a net outflow of $7,280. From Year 2 onward the app saves $2,600 every year. The 3-Year total nets these against each other.
Finding
Optellum Lung AI already has a vendor-hosted SaaS offering (Optellum Cloud) that would eliminate the two on-premises Windows Server VMs currently running in MUSC-USB. The transition is strategically sound but cannot be executed until a HIPAA Business Associate Agreement is confirmed with Optellum Cloud and the existing contract terms are reviewed. Once those conditions are satisfied, this is a straightforward vendor-managed SaaS migration with an estimated $2,600 annual savings.
In plain terms: this app should move to the vendor's cloud product, but a few specific items must be confirmed first. See Conditions to Unlock below.
Conditions to Unlock (2 items)
This recommendation becomes executable once these conditions are confirmed:
☐ Confirm HIPAA Business Associate Agreement with Optellum before any cloud migration
Owner: Compliance Officer / Legal · Timeline: 10 days · Unlocks: compliance_close, cap_resolution, migration_commit
☐ Pull the current Optellum contract and get SaaS pricing before committing to migration
Owner: Procurement / Vendor Management · Timeline: 15 days · Unlocks: cost_clarity, migration_commit, cap_resolution
See Backlog tab for full action backlog and P2/P3 items.
This is the list of work needed to move this recommendation forward. The 2 P1 items are the priority — they unlock the recommendation. P2 and P3 items improve confidence and reduce risk but are not blockers. 'Story points' measure effort, not time or cost — see the Methodology tab for how they work.
55%
Current
60%
Threshold
62%
Ceiling
25
Story Points
Why This Score
Current confidence of 55 is bound by SRC-CAP-002 (majority of 7 evidence dimensions are INFERRED or MISSING: procurement_contracts=MISSING, architecture_diagrams=MISSING, dr_requirements=MISSING, dependency_mapping=INFERRED, security_compliance=INFERRED), which imposes a hard ceiling of 55 regardless of pre-cap score (61.75). The backlog is structured to resolve the three MISSING dimensions and upgrade the two INFERRED dimensions to CONFIRMED, which is the only path to lifting SRC-CAP-002 and allowing confidence to climb above 55 in a subsequent run.
Proceeding with Optellum Cloud SaaS migration without completing APP-032-001 (BAA) and APP-032-002 (contract review) would constitute a HIPAA violation risk and an unquantified contractual commitment — both conditions must be satisfied before any PHI is transmitted to the vendor-hosted environment.
Action Backlog — 6 Items · 25 Story Points · Ceiling: 62%
APP-032-001 Execute HIPAA BAA with Optellum for Optellum Cloud SaaS
P1 BAA/Legal
Obtain and execute a HIPAA-compliant Business Associate Agreement (BAA) with Optellum covering MUSC Health as the contracting entity for Optellum Cloud. The catalog field baa_valid is null, resolving to baa_confirmed=UNKNOWN under v3.1 tri-state rules. phi_handling=true means no PHI may be transmitted to Optellum Cloud until a BAA is in place. This is the single most critical unlock condition: wit
+0% confidence -0% ROI variance 👤 📋 · 3 pts
APP-032-002 Obtain and Review Existing Optellum Contract Terms and SaaS Migration Pricing
P1 Procurement
Retrieve the existing on-premises Optellum license agreement and engage Optellum's account team to obtain: (1) current contract ACV and renewal/termination date, (2) SaaS migration pricing for Optellum Cloud, (3) confirmation of FDA clearance continuity under the cloud-hosted deployment model, and (4) any data residency or HIPAA addendum terms specific to Optellum Cloud. The procurement_contracts
+7% confidence -0% ROI variance 👤 📋 · 5 pts
APP-032-003 Validate PACS/DICOM Integration Compatibility with Optellum Cloud SaaS Endpoint
P2 Architecture
Engage the Radiology informatics team (Wendy Ketchum) and MUSC network/integration team to produce an integration architecture diagram documenting: (1) the PACS-to-Optellum DICOM feed path (current on-premises and proposed SaaS endpoint), (2) any HL7 or RIS order feed dependencies, (3) downstream result routing to EHR or radiologist workstation, and (4) network path requirements (Direct Connect or
+7% confidence -0% ROI variance 👤 📋 · 5 pts
APP-032-004 Normalize phi_handled Field and Complete Security/Compliance Classification
P2 Security/Compliance
Resolve two related catalog gaps: (1) normalize the catalog field from 'phi_handling' to the canonical 'phi_handled' and obtain a compliance officer attestation confirming phi_handled=true with an artifact citation, and (2) complete a HIPAA/HITRUST security classification review for Optellum Lung AI to confirm the PHI handling scope, data elements processed (DICOM patient demographics, accession n
+7% confidence -0% ROI variance 👤 📋 · 3 pts
APP-032-005 Deploy 30-Day APM Telemetry Collection on optellsql and optelltstsql
P2 Telemetry
Deploy an APM agent (e.g., Datadog, Dynatrace, or Azure Monitor) on both Optellum VMs (optellsql, optelltstsql) to collect 30+ days of CPU and RAM time-series data. Current telemetry is a single RVTools point-in-time snapshot with cpu_avg_pct null on both VMs, triggering the SRC-RS-002 hard cap (MODERATE, 25% reduction) and SRC-CAP-004 (ceiling 65). A 30-day APM dataset will: (1) upgrade telemetry
+0% confidence -0% ROI variance 👤 📋 · 8 pts
APP-032-006 Resolve Life Safety Classification and Document DR Requirements
P3 Architecture
Two related MEDIUM-severity gaps addressed in a single lightweight review: (1) Engage clinical informatics and the Radiology team (Wendy Ketchum) to formally classify whether Optellum Lung AI meets the v3.1 life_safety_classification criteria. The application is diagnostic decision support (not treatment delivery), making life_safety=false likely, but the catalog field is absent and cannot be conf
+7% confidence -0% ROI variance 👤 📋 · 1 pts
This tab shows the evidence behind the recommendation. The seven dimensions are scored by quality — CONFIRMED evidence counts fully, INFERRED evidence counts less. The validator table confirms each analysis step passed its checks. Green means the step is sound.
Evidence Classification — 7 Dimensions
Dimension Classification Multiplier Contribution
Telemetry SINGLE 0.50 6.00
Dependency Mapping INFERRED 0.25 2.50
Procurement Contracts MISSING 0.00 0.00
Architecture Diagrams MISSING 0.00 0.00
Security Compliance INFERRED 0.25 1.25
Business Criticality SINGLE 0.50 2.00
Dr Requirements MISSING 0.00 0.00
Applied Cap: SRC-CAP-002
Validation Reports — Per Agent
Agent Status Provenance Counts Defaults Used Rule Failures
telemetry PASS
SRC:2CUST:8COMP:5DEF:2
 None ✓ None
dependency PASS
SRC:2CUST:6COMP:5DEF:4
 None ✓ None
procurement PASS
None ✓ None
provisioning PASS_WITH_DEFAULTS
SRC:7CUST:1COMP:9DEF:1
• {'default_id': 'DEF-WL-001', 'field': 'workload_type', 'value': 'app_server', 'r
• {'default_id': 'DEF-MIG-001', 'field': 'migration_factor', 'value': 0.2, 'reason
 ✓ None
synthesizer PASS
None ✓ None
confidence_advisor PASS
COMP:18
 None ✓ None
This section contains the full technical reasoning behind the recommendation: the rules that fired, the decision path, the confidence math, and the open intelligence gaps. It is intended for analysts and engineers. Executive readers can rely on the Overview tab.
Rules Applied
These are the specific governance rules that produced this recommendation.
▸ RULE 1 — REPURCHASE
A vendor SaaS replacement is available and credible, and the app has no hardware constraint — so the recommendation is to replace it with the vendor's cloud service.
▸ SRC-CAP-002 (binding)
Confidence is capped at 55 because more than half of the seven evidence dimensions are inferred or missing — the conclusion is data-thin.
▸ SRC-CAP-003
Confidence is capped at 65 because none of the seven evidence dimensions reached CONFIRMED quality — there is no high-quality anchor for the analysis.
▸ SRC-CAP-004
Confidence is capped at 65 because telemetry is a single point-in-time snapshot with no ongoing monitoring — right-sizing needs usage patterns over time.
▸ RULE T2 — Right-Sizing Hard Cap
Right-sizing is limited to a moderate 25% reduction because telemetry is a single snapshot. More aggressive right-sizing needs 30+ days of monitoring data.
Decision Path
The step-by-step trace of how the recommendation was reached, in the system's own terms.
PRE-CHECK: hardware_dependency=false, life_safety_classification=UNKNOWN → structural blocker does not fire. RULE 2 (RETIRE): no decommission annotation, no named successor → does not fire. RULE 1 (REPURCHASE): saas_alternative_available=true (catalog.notes), saas_confidence_inferred=72 ≥ 50, hardware_dependency=false → FIRES. strategic_recommendation=REPURCHASE. CONFIDENCE: base=50, telemetry(SINGLE)=+6.0, dependency(INFERRED)=+2.5, procurement(MISSING)=+0.0, architecture(MISSING)=+0.0, security(INFERRED)=+1.25, criticality(SINGLE)=+2.0, dr(MISSING)=+0.0 → pre-cap=61.75. CAP EVALUATION: SRC-CAP-002 (5/7 INFERRED/MISSING)=55, SRC-CAP-003 (0 CONFIRMED)=65, SRC-CAP-004 (snapshot=1, no APM)=65 → lowest=SRC-CAP-002 → final_confidence=55. GATE (v4.1): final_confidence=55 < 60 AND strategic_recommendation=REPURCHASE → recommendation_status=PENDING_CONDITIONS, displayed_primary=REPURCHASE (gate does NOT force RETAIN in v4.1). CONDITIONS TO UNLOCK: (1) Confirm HIPAA BAA with Optellum Cloud, (2) Obtain and review existing Optellum contract terms, (3) Validate PACS/DICOM integration compatibility with SaaS endpoint, (4) Confirm phi_handled classification via security/compliance review.
Cap Evaluation (all fired caps): SRC-CAP-002: ≥4 of 7 dimensions INFERRED/MISSING → ceiling 55 (ceiling=55) ← BINDING SRC-CAP-003: Zero CONFIRMED dimensions → ceiling 65 (ceiling=65) SRC-CAP-004: Single snapshot, no APM → ceiling 65 (ceiling=65)
Confidence Scoring Breakdown
How the confidence percentage was calculated, dimension by dimension.
Base score: 50 Evidence contributions: Telemetry SINGLE +6.00 Dependency Mapping INFERRED +2.50 Procurement Contracts MISSING +0.00 Architecture Diagrams MISSING +0.00 Security Compliance INFERRED +1.25 Business Criticality SINGLE +2.00 Dr Requirements MISSING +0.00 Pre-cap score: 61.75 Binding cap applied: SRC-CAP-002 Final confidence: 55%
Intelligence Gaps
Specific things we don't yet know. Each gap is a reason the confidence score is held back.
BAA with Optellum Cloud unconfirmed
phi_handling=true in catalog; baa_confirmed=UNKNOWN blocks HIPAA-compliant SaaS migration. Must execute BAA before any PHI is transmitted to Optellum Cloud. Resolves SRC-CAP-001 risk and is the primary unlock condition.
Procurement contract terms unavailable
contract_intelligence=INSUFFICIENT_EVIDENCE; existing on-premises license terms, termination clauses, and SaaS migration pricing are unknown. Required before REPURCHASE can be executed.
PACS/DICOM integration compatibility with Optellum Cloud unvalidated
Upstream PACS dependency is inferred but unconfirmed. SaaS migration requires validation that Optellum Cloud can receive DICOM feeds from MUSC's PACS infrastructure without on-premises intermediary VMs.
phi_handled field unresolved (UNKNOWN)
Catalog phi_handling=true but dependency agent resolved phi_handled=UNKNOWN due to absent confirmation artifact. Security/compliance review required to formally classify PHI handling scope and confirm HIPAA applicability to Optellum Cloud.
No APM telemetry; single RVTools snapshot
Utilization profile based on single snapshot (avg_mem_active_pct=12%). Cannot confirm workload patterns, peak load behavior, or AI inference latency requirements. Limits right-sizing confidence and migration sizing.
life_safety_classification unresolved (UNKNOWN)
Lung AI supports clinical decision-making for nodule detection. If classified as life-safety, migration risk escalates and additional clinical validation may be required. Requires clinical informatics review.
DR requirements undocumented
No DR documentation available. RTO/RPO requirements for a High-criticality radiology AI platform are unknown. SaaS SLA from Optellum Cloud must be validated against MUSC DR standards.
RIS/EHR order feed integration unconfirmed
Upstream RIS or HL7 order feed is inferred from industry pattern. epic_integration=false rules out Epic but does not confirm or eliminate other HL7 feeds. Integration inventory required before cutover.
Technical Detail
A fuller technical explanation of the recommendation and its constraints.
Procurement agent confirmed saas_available=true with saas_confidence_inferred=72 and explicitly identified Optellum Cloud as the vendor-hosted SaaS alternative, satisfying all three conditions for Rule 1 (REPURCHASE): SaaS available, inferred confidence ≥ 50, and hardware_dependency=false per dependency agent. The pre-check structural blocker does not apply as hardware_dependency=false and life_safety_classification=UNKNOWN (not confirmed true). Confidence is capped at 55 by SRC-CAP-002 (5 of 7 dimensions INFERRED or MISSING) and SRC-CAP-004 (single snapshot, no APM telemetry), placing the recommendation in PENDING_CONDITIONS status. The binding unlock condition is BAA confirmation with Optellum Cloud and procurement contract review.
Secondary Path
The fallback option if the primary recommendation cannot proceed.
Secondary Path — REHOST
REHOST is the secondary disposition if SaaS migration is blocked: both VMs are Windows Server 2022 with no hardware dependency, making a standard IaaS lift-and-shift viable. However, given the vendor's own SaaS offering, REHOST would be a suboptimal interim state.
Right-Sizing Notes
Why the cloud sizing estimate is set where it is, based on the telemetry available.
Fleet avg mem_active_pct = 12.0% across 2 powered-on VMs (optellsql: 8.0%, optelltstsql: 16.0%). Raw utilization signal would classify as OVERSIZED (SRC-RS-003) under Rule T3 (threshold < 25%). However, Rule T2 is active: snapshot_count=1 and apm_telemetry_available=false. A single point-in-time snapshot cannot distinguish a genuinely idle VM from one sampled between burst cycles. Tier is hard-capped at SRC-RS-002 (MODERATE, 0.25 reduction); SRC-RS-003 and SRC-RS-004 are forbidden. CPU utilization data is entirely unavailable (cpu_avg_pct null on all 2 VMs; no APM supplement). No EOL OS violations detected — both VMs run Windows Server 2022 (supported). phi_handling=true with baa_valid=null; BAA confirmation gap is flagged for synthesizer attention. Application type is AI/Cloud-Connected with a noted SaaS alternative (Optellum Cloud); disposition analysis deferred to synthesizer.
This tab shows the raw output of each analysis agent — the specialized components that examine telemetry, dependencies, contracts, and cost. It is the most detailed and most technical view. Most readers will find the Overview and Backlog tabs sufficient.
Telemetry Agent
Utilization Profilemoderate
Snapshot Count1
APM AvailableFalse
Hard Rule TriggeredTrue
Right-Sizing Opp.True
Agent Confidence45%
Telemetry — VM Counts
Powered Off VMs0
Flagged VMs0
Mem Active % (avg)12.0%
Data WindowNone
Inventory Source
Dependency Agent
Epic IntegrationFalse
PHI HandledUNKNOWN
Hardware DependencyFalse
Latency SensitiveFalse
Blast Radiusmedium
Migration RiskMEDIUM
Conf (Inferred)72%
Conf (Confirmed)38%
Verificationsingle_source
Cloud Connectivity
ComplianceHIPAA / HITRUST
Upstream (4)
PACS / Radiology Imaging System: inferred: Optellum Lung AI is a CT imaging analysis platform; DICOM image feed from PACS is a structural requirement of…
RIS / Radiology Information System or EHR Order Feed: inferred: Lung nodule AI platforms typically receive worklist or order context from RIS or EHR; catalog.epic_integration…
Active Directory / Windows Authentication: inferred: Both VMs run Windows Server 2022; domain authentication is standard for clinical application VMs in…
Optellum Cloud (vendor SaaS / AI inference endpoint): confirmed: catalog.notes explicitly states 'SaaS alternative: Optellum Cloud (vendor-hosted SaaS)';…
Downstream (2)
Radiologist Workstation / Reporting Workflow: inferred: Lung AI output (nodule detection, risk scores) is consumed by radiologists for reporting; Radiology team…
EHR / Clinical Documentation System (results delivery): inferred: AI-generated findings are typically surfaced in the EHR or RIS for clinical action; catalog.epic_integration=f…
Procurement Agent
SaaS Available—
SaaS VendorOptellum
SaaS ProductOptellum Cloud
Repurchase ViableINSUFFICIENT_EVIDENCE
BAA ConfirmedUNKNOWN
SaaS Confidence
Inferred Score72%
Confirmed Score0%
Verificationsingle_source
AWS Equiv.None
Provisioning Agent (Python engine — not LLM)
Current VMs2
Recommended VMs2
Right-Sizing25% reduction
Current TCO/yr$13,000
Projected Cloud TCO$10,400
Migration One-Time$2,080
Dual-Running Cost$7,800
-$7,280
Year-1 Net
$2,600
Steady-State
-$2,080
3-Year
AWS TargetAmazon EC2
Azure TargetAzure Virtual Machines
Cost Modeldefault_multipliers
DEF: {'default_id': 'DEF-WL-001', 'field': 'workload_type', 'value': 'app_server', 'reason': 'workload_type indeterminable'}
DEF: {'default_id': 'DEF-MIG-001', 'field': 'migration_factor', 'value': 0.2, 'reason': 'migration_complexity not provided'}
DEF: {'default_id': 'DEF-COST-001', 'field': 'cost_model_basis', 'value': 'default_multipliers', 'reason': 'no customer cloud quote provided; using SRC-TCO-* and SRC-MULT-* registry defaults'}
cNeuro
Vendor: Cortechs.ai · 1 VMs · Criticality: High · HIPAA / HITRUST — BAA status unconfirmed
REHOST
55%
Confidence
28%
Data Quality
ROI: Year-1 Net: -$3,640 · Steady-State: $1,300/yr · 3-Year: -$1,040
Overview
Backlog
Provenance
Technical Details
Agents
Recommendation
REHOST ⏳ PENDING CONDITIONS
SRC-CAP-002 Binding cap · Ceiling: 55% · Threshold: 60%
Secondary: RETAIN (pending BAA and PHI confirmation)
Confidence
55%
Viable, but specific gaps must close first.
ROI Summary (Python-computed, not LLM)
-$3,640
Year-1 Net
$1,300
Steady-State /yr
-$1,040
3-Year Total
ROI variance: ±60%
⚙ Financials computed by provisioning_engine_v3 (Python) — not LLM-generated
In plain terms: migrating costs money up front — Year 1 is a net outflow of $3,640. From Year 2 onward the app saves $1,300 every year. The 3-Year total nets these against each other.
Finding
cNeuro is a single-VM Windows Server 2019 workload from Cortechs.ai running neuroimaging AI analysis at MUSC-USB. The application has no confirmed hardware dependency and is a candidate for standard cloud rehosting (IaaS lift-and-shift). However, the BAA status is unconfirmed and PHI handling is undocumented, which must be resolved before migration can proceed given the HIPAA/HITRUST compliance scope.
In plain terms: see the recommendation and conditions above.
Conditions to Unlock (3 items)
This recommendation becomes executable once these conditions are confirmed:
☐ Verify Business Associate Agreement for cNeuro
Owner: Compliance Officer / Legal Counsel · Timeline: 14 days · Unlocks: compliance_close, cap_resolution, migration_commit
☐ Confirm Whether cNeuro Handles Patient Data and Whether It Is Live in Production
Owner: Application Owner / HIPAA Privacy Officer · Timeline: 10 days · Unlocks: compliance_close, cap_resolution, migration_commit
☐ Install Application Performance Monitoring on cNeuro Server
Owner: Infrastructure Engineer / Cloud Migration Team · Timeline: 35 days · Unlocks: cap_resolution, cost_clarity, migration_commit
See Backlog tab for full action backlog and P2/P3 items.
This is the list of work needed to move this recommendation forward. The 3 P1 items are the priority — they unlock the recommendation. P2 and P3 items improve confidence and reduce risk but are not blockers. 'Story points' measure effort, not time or cost — see the Methodology tab for how they work.
55%
Current
60%
Threshold
64%
Ceiling
27
Story Points
Why This Score
cNeuro is capped at 55 by SRC-CAP-002 because 5 of 7 evidence dimensions are INFERRED or MISSING — specifically: dependency_mapping (INFERRED), procurement_contracts (MISSING), architecture_diagrams (INFERRED), security_compliance (INFERRED), and dr_requirements (MISSING). The pre-cap confidence of 63.5 reflects a structurally viable REHOST candidate (single VM, no hardware dependency, Windows Server 2019), but the binding cap correctly reflects that no dimension has been independently confirmed, and the BAA/PHI compliance posture for this HIPAA-scoped neuroimaging workload remains unresolved.
Proceeding to cloud migration without resolving BAA confirmation and PHI handling classification exposes the organization to HIPAA enforcement risk and potential OCR audit findings for unauthorized PHI transmission to a cloud environment without a signed Business Associate Agreement.
Action Backlog — 6 Items · 27 Story Points · Ceiling: 64%
APP-033-001 Confirm BAA Status with Cortechs.ai and Target Cloud Provider
P1 BAA/Legal
Retrieve or execute a signed Business Associate Agreement (BAA) with Cortechs.ai covering the on-premises deployment, and confirm BAA availability with the target cloud provider (AWS, Azure, GCP, or OCI). Update catalog field baa_valid with the confirmed value and attach the BAA document reference. This is the single highest-priority compliance blocker: phi_handling=true in the catalog and the dep
+0% confidence -0% ROI variance 👤 📋 · 3 pts
APP-033-002 Classify PHI Handling Scope and Confirm Production vs. POC Status
P1 Security/Compliance
Conduct a structured interview with the application owner (Wendy Ketchum, USB) and the HIPAA Privacy Officer to: (1) confirm whether cNeuro processes, stores, or transmits PHI (DICOM studies with embedded patient demographics constitute PHI under HIPAA); (2) resolve the governance discrepancy between catalog.environment=Production and the VM note reading 'cNeuro POC' — determine whether this is an
+9% confidence -0% ROI variance 👤 📋 · 3 pts
APP-033-003 Deploy 30-Day APM Telemetry Collection on cneuroprd
P1 Telemetry
Install a lightweight APM agent (e.g., AWS CloudWatch Agent, Azure Monitor Agent, Datadog, or equivalent) on the cneuroprd VM to collect CPU utilization, memory active percentage, disk I/O, and network throughput over a minimum 30-day window covering representative DICOM processing load. Current telemetry is a single RVTools snapshot (snapshot_count=1, apm_telemetry_available=false), which trigger
+0% confidence -0% ROI variance 👤 📋 · 8 pts
APP-033-004 Obtain Cortechs.ai Contract and Assess SaaS / Cloud Delivery Options
P2 Vendor Engagement
Engage Cortechs.ai account management to: (1) retrieve the current license agreement, ACV, and renewal date for the cNeuro deployment at MUSC-USB; (2) determine whether Cortechs.ai offers a cloud-hosted or SaaS delivery model for cNeuro (saas_available=UNKNOWN, saas_confidence_inferred=35); (3) confirm whether the existing license is portable to a cloud-hosted VM (IaaS REHOST path) without additio
+0% confidence -0% ROI variance 👤 📋 · 5 pts
APP-033-005 Document DR / RTO-RPO Requirements for cNeuro
P2 Architecture
Work with the clinical application owner and IT leadership to document formal DR requirements for cNeuro: RTO (Recovery Time Objective), RPO (Recovery Point Objective), and backup frequency. The application is classified High criticality and is a single-VM deployment with no HA or DR topology visible (dr_requirements=MISSING). For a HIPAA-scoped neuroimaging AI workload, the absence of documented
+9% confidence -0% ROI variance 👤 📋 · 3 pts
APP-033-006 Confirm DICOM / PACS Integration Topology and Network Requirements
P2 Dependency Mapping
Conduct network-level dependency mapping to confirm the DICOM/PACS upstream integration for cNeuro: (1) identify the specific PACS system(s) or modality worklist feeds that send DICOM studies to cneuroprd; (2) document port, protocol (DICOM C-STORE, C-FIND, or DIMSE), and data volume; (3) confirm whether Cortechs.ai cloud-assist processing requires outbound connectivity to vendor cloud endpoints (
+0% confidence -0% ROI variance 👤 📋 · 5 pts
This tab shows the evidence behind the recommendation. The seven dimensions are scored by quality — CONFIRMED evidence counts fully, INFERRED evidence counts less. The validator table confirms each analysis step passed its checks. Green means the step is sound.
Evidence Classification — 7 Dimensions
Dimension Classification Multiplier Contribution
Telemetry SINGLE 0.50 6.00
Dependency Mapping INFERRED 0.25 2.50
Procurement Contracts MISSING 0.00 0.00
Architecture Diagrams INFERRED 0.25 1.75
Security Compliance INFERRED 0.25 1.25
Business Criticality SINGLE 0.50 2.00
Dr Requirements MISSING 0.00 0.00
Applied Cap: SRC-CAP-002
Validation Reports — Per Agent
Agent Status Provenance Counts Defaults Used Rule Failures
telemetry PASS
SRC:2CUST:9COMP:3DEF:2
 None ✓ None
dependency PASS
SRC:3CUST:1COMP:5DEF:3
 None ✓ None
procurement PASS
None ✓ None
provisioning PASS_WITH_DEFAULTS
SRC:7CUST:1COMP:9DEF:1
• {'default_id': 'DEF-WL-001', 'field': 'workload_type', 'value': 'app_server', 'r
• {'default_id': 'DEF-MIG-001', 'field': 'migration_factor', 'value': 0.2, 'reason
 ✓ None
synthesizer PASS
None ✓ None
confidence_advisor PASS
COMP:10
 None ✓ None
This section contains the full technical reasoning behind the recommendation: the rules that fired, the decision path, the confidence math, and the open intelligence gaps. It is intended for analysts and engineers. Executive readers can rely on the Overview tab.
Rules Applied
These are the specific governance rules that produced this recommendation.
▸ RULE 6 — REHOST
No other qualifying path applies — the default for virtual workloads is to rehost to cloud IaaS.
▸ SRC-CAP-002 (binding)
Confidence is capped at 55 because more than half of the seven evidence dimensions are inferred or missing — the conclusion is data-thin.
▸ SRC-CAP-003
Confidence is capped at 65 because none of the seven evidence dimensions reached CONFIRMED quality — there is no high-quality anchor for the analysis.
▸ SRC-CAP-004
Confidence is capped at 65 because telemetry is a single point-in-time snapshot with no ongoing monitoring — right-sizing needs usage patterns over time.
▸ RULE T2 — Right-Sizing Hard Cap
Right-sizing is limited to a moderate 25% reduction because telemetry is a single snapshot. More aggressive right-sizing needs 30+ days of monitoring data.
Decision Path
The step-by-step trace of how the recommendation was reached, in the system's own terms.
PRE-CHECK: hardware_dependency=false, life_safety_classification=UNKNOWN — structural blocker does not fire. RULE 1 (REPURCHASE): saas_alternative_available=UNKNOWN and saas_confidence_inferred=35 < 50 — does not fire. RULE 2 (RETIRE): no redundancy_identified, no decommission_annotation_present — does not fire. RULE 3 (REFACTOR): vendor COTS application, not internally developed — does not fire. RULE 4 (REHOST): VM-based workload confirmed (1 VM, poweredOn), hardware_dependency=false, no confirmed latency constraint — FIRES. strategic_recommendation=REHOST. CONFIDENCE GATE: final_confidence=55 < 60, strategic_recommendation=REHOST (not RETAIN) → recommendation_status=PENDING_CONDITIONS, displayed_primary=REHOST (gate does not force RETAIN under v4.1). CAPS APPLIED: SRC-CAP-002 (5/7 dimensions INFERRED/MISSING, cap 55) is binding; SRC-CAP-003 (zero CONFIRMED, cap 65) and SRC-CAP-004 (snapshot_count=1 + no APM, cap 65) also apply but are non-binding. Pre-cap total 63.5 reduced to 55.
Cap Evaluation (all fired caps): SRC-CAP-002: ≥4 of 7 dimensions INFERRED/MISSING → ceiling 55 (ceiling=55) ← BINDING SRC-CAP-003: Zero CONFIRMED dimensions → ceiling 65 (ceiling=65) SRC-CAP-004: Single snapshot, no APM → ceiling 65 (ceiling=65)
Confidence Scoring Breakdown
How the confidence percentage was calculated, dimension by dimension.
Base score: 50 Evidence contributions: Telemetry SINGLE +6.00 Dependency Mapping INFERRED +2.50 Procurement Contracts MISSING +0.00 Architecture Diagrams INFERRED +1.75 Security Compliance INFERRED +1.25 Business Criticality SINGLE +2.00 Dr Requirements MISSING +0.00 Pre-cap score: 63.5 Binding cap applied: SRC-CAP-002 Final confidence: 55%
Intelligence Gaps
Specific things we don't yet know. Each gap is a reason the confidence score is held back.
BAA Confirmation Required
baa_confirmed=UNKNOWN for a HIPAA/HITRUST-scoped application handling potential PHI; no cloud migration can proceed without a signed BAA with the target cloud provider. This is the single highest-priority blocker.
PHI Handling Classification
phi_handled=UNKNOWN despite catalog phi_handling=true; the dependency agent could not confirm PHI scope. Neuroimaging studies (DICOM/PACS) may contain PHI. Must be confirmed before migration architecture is finalized.
Procurement Contract Intelligence
contract_intelligence=INSUFFICIENT_EVIDENCE; no vendor contract data available. Cannot assess license portability to cloud, SaaS upgrade path availability, or migration cost obligations without contract review.
SaaS Alternative Assessment
saas_available=UNKNOWN; Cortechs.ai may offer a cloud-hosted SaaS version of cNeuro. If confirmed, REPURCHASE would be preferred over REHOST. Vendor engagement required to assess cloud product roadmap.
APM / Application Telemetry
apm_telemetry_available=false; only RVTools snapshot available. Application-layer performance, transaction volumes, and DICOM processing throughput are unknown. Required for right-sizing validation and migration risk assessment.
DR / RTO-RPO Requirements
No DR requirements documented for a High-criticality application. Single-VM deployment with no redundancy; RTO/RPO targets unknown. Must be defined before cloud architecture is specified.
DICOM / PACS Integration Confirmation
Upstream DICOM/PACS dependency is inferred from vendor product function, not confirmed via network mapping or integration documentation. PACS connectivity requirements (port, protocol, latency) must be validated before migration.
POC vs. Production Status Clarification
VM note reads 'cNeuro POC Requested by Wendy Ketchum USB Windows Server 2019' — it is unclear whether this is a production workload or an active proof-of-concept. If POC, disposition may shift to RETIRE or RETAIN pending formal production decision.
Technical Detail
A fuller technical explanation of the recommendation and its constraints.
Telemetry confirms a single powered-on VM (cneuroprd, 4 vCPU / 16 GiB RAM) with moderate utilization (8% avg memory active) and no hardware dependency flagged by the dependency agent. Procurement confirms no SaaS alternative is currently identified (saas_confidence_inferred=35, below the 50 threshold for Rule 1), ruling out REPURCHASE. No decommission annotation or named successor exists, ruling out RETIRE. The application is vendor COTS (Cortechs.ai), ruling out REFACTOR. Rule 4 (REHOST) fires: VM-based, no hardware dependency, no confirmed latency constraint. Confidence is capped at 55 by SRC-CAP-002 (5 of 7 dimensions INFERRED or MISSING), triggering PENDING_CONDITIONS status; the binding gaps are BAA confirmation, PHI handling classification, and procurement contract intelligence.
Secondary Path
The fallback option if the primary recommendation cannot proceed.
Secondary Path — RETAIN (pending BAA and PHI confirmation)
If discovery confirms a Cortechs.ai cloud-hosted SaaS offering with adequate PHI controls and BAA coverage, REPURCHASE should be re-evaluated as the preferred disposition, eliminating the on-premises VM footprint entirely.
Right-Sizing Notes
Why the cloud sizing estimate is set where it is, based on the telemetry available.
Single RVTools snapshot (snapshot_count=1, apm_telemetry_available=false). Rule T2 hard cap applied: maximum tier is SRC-RS-002 (MODERATE, 0.25 reduction). Observed fleet_avg_mem_active_pct=8.0% would suggest OVERSIZED under Rule T3, but that inference is explicitly forbidden with single-snapshot data — a VM sampled between bursts is indistinguishable from a genuinely idle VM. cpu_avg_pct is null on all VMs; no CPU-side sizing evidence available. VM note indicates 'cNeuro POC' status — if this VM is still in proof-of-concept phase, workload patterns may not yet be representative of steady-state production load, further supporting the conservative MODERATE cap. Right-sizing opportunity flagged as true at MODERATE tier pending deeper telemetry collection (minimum 30-day APM or multi-snapshot RVTools series).
This tab shows the raw output of each analysis agent — the specialized components that examine telemetry, dependencies, contracts, and cost. It is the most detailed and most technical view. Most readers will find the Overview and Backlog tabs sufficient.
Telemetry Agent
Utilization Profilemoderate
Snapshot Count1
APM AvailableFalse
Hard Rule TriggeredTrue
Right-Sizing Opp.True
Agent Confidence35%
Telemetry — VM Counts
Powered Off VMs0
Flagged VMs0
Mem Active % (avg)8.0%
Data WindowNone
Inventory Source
Dependency Agent
Epic IntegrationFalse
PHI HandledUNKNOWN
Hardware DependencyFalse
Latency SensitiveUNKNOWN
Blast Radiusmedium
Migration RiskMEDIUM
Conf (Inferred)62%
Conf (Confirmed)28%
Verificationsingle_source
Cloud Connectivity
ComplianceHIPAA / HITRUST
Upstream (3)
DICOM / PACS (Picture Archiving and Communication System): inferred: vendor product function — cNeuro ingests neuroimaging DICOM studies; a PACS or modality worklist feed is…
Active Directory / Windows Authentication: inferred: application_type=Client/Server on Windows Server 2019; standard enterprise Windows authentication dependency…
Cortechs.ai Cloud Analysis Service (optional/hybrid): inferred: Cortechs.ai product architecture supports optional cloud-assisted processing; on-premise deployment (single…
Downstream (2)
Radiology / Neurology Reporting Workflow: inferred: cNeuro produces quantitative volumetric reports consumed by radiologists or neurologists; downstream…
Epic EHR (indirect / report delivery): inferred: catalog.epic_integration=false (confirmed false); however, report output may be delivered to Epic via HL7 or…
Procurement Agent
SaaS Available—
SaaS VendorCortechs.ai
SaaS Product
Repurchase ViableUNKNOWN
BAA ConfirmedUNKNOWN
SaaS Confidence
Inferred Score35%
Confirmed Score0%
Verificationinferred_only
AWS Equiv.None
Provisioning Agent (Python engine — not LLM)
Current VMs1
Recommended VMs1
Right-Sizing25% reduction
Current TCO/yr$6,500
Projected Cloud TCO$5,200
Migration One-Time$1,040
Dual-Running Cost$3,900
-$3,640
Year-1 Net
$1,300
Steady-State
-$1,040
3-Year
AWS TargetAmazon EC2
Azure TargetAzure Virtual Machines
Cost Modeldefault_multipliers
DEF: {'default_id': 'DEF-WL-001', 'field': 'workload_type', 'value': 'app_server', 'reason': 'workload_type indeterminable'}
DEF: {'default_id': 'DEF-MIG-001', 'field': 'migration_factor', 'value': 0.2, 'reason': 'migration_complexity not provided'}
DEF: {'default_id': 'DEF-COST-001', 'field': 'cost_model_basis', 'value': 'default_multipliers', 'reason': 'no customer cloud quote provided; using SRC-TCO-* and SRC-MULT-* registry defaults'}
Volpara
Vendor: Volpara Health · 1 VMs · Criticality: High · HIPAA / HITRUST — PHI handled; BAA requi
REPURCHASE
40%
Confidence
28%
Data Quality
ROI: Year-1 Net: -$3,640 · Steady-State: $1,300/yr · 3-Year: -$1,040
Overview
Backlog
Provenance
Technical Details
Agents
Recommendation
REPURCHASE ⏳ PENDING CONDITIONS
SRC-CAP-001 Binding cap · Ceiling: 40% · Threshold: 60%
Secondary: RETAIN
Confidence
40%
Viable, but specific gaps must close first.
ROI Summary (Python-computed, not LLM)
-$3,640
Year-1 Net
$1,300
Steady-State /yr
-$1,040
3-Year Total
ROI variance: ±60%
⚙ Financials computed by provisioning_engine_v3 (Python) — not LLM-generated
In plain terms: migrating costs money up front — Year 1 is a net outflow of $3,640. From Year 2 onward the app saves $1,300 every year. The 3-Year total nets these against each other.
Finding
Volpara Health offers a vendor-hosted SaaS version of this mammography analytics platform, making a direct repurchase the preferred path to eliminate the on-premises virtual appliance. However, because this system handles patient imaging data under HIPAA, a Business Associate Agreement with Volpara Health for their cloud product must be confirmed before migration can proceed. Once BAA status, contract terms, and integration dependencies (DICOM feed, PACS, RIS) are validated, this workload is a strong candidate for SaaS transition.
In plain terms: this app should move to the vendor's cloud product, but a few specific items must be confirmed first. See Conditions to Unlock below.
Conditions to Unlock (2 items)
This recommendation becomes executable once these conditions are confirmed:
☐ Verify Business Associate Agreement for Volpara Cloud Migration
Owner: Compliance Officer / Legal · Timeline: 10 days · Unlocks: compliance_close, cap_resolution, migration_commit
☐ Secure SaaS Pricing and Contract Terms from Volpara Health
Owner: Procurement / Vendor Management · Timeline: 21 days · Unlocks: cost_clarity, migration_commit
See Backlog tab for full action backlog and P2/P3 items.
This is the list of work needed to move this recommendation forward. The 2 P1 items are the priority — they unlock the recommendation. P2 and P3 items improve confidence and reduce risk but are not blockers. 'Story points' measure effort, not time or cost — see the Methodology tab for how they work.
40%
Current
60%
Threshold
55%
Ceiling
20
Story Points
Why This Score
Current confidence of 40 is the direct result of SRC-CAP-001 (phi_handled=true, baa_confirmed=UNKNOWN), which overrides the pre-cap calculated score of 63.5 and establishes a hard ceiling at 40 — the same value as the current score, leaving zero available headroom for confidence improvement in this run. The pre-cap score of 63.5 reflects a well-understood workload with a clear SaaS path, but the absence of BAA confirmation for a PHI-handling mammography analytics platform is a non-negotiable compliance gate that prevents any confidence climb until resolved.
Proceeding with any data migration or SaaS transition to Volpara Health Cloud before BAA confirmation exposes the organization to HIPAA violation risk and potential OCR enforcement action for unauthorized PHI disclosure to a vendor without an executed Business Associate Agreement.
Action Backlog — 6 Items · 20 Story Points · Ceiling: 55%
APP-034-001 Confirm BAA with Volpara Health for Volpara Health Cloud SaaS product
P1 BAA/Legal
Engage the compliance officer and Volpara Health account team to determine whether an executed BAA exists covering Volpara Health Cloud (vendor-hosted SaaS) for this organization as a HIPAA covered entity. If a BAA exists, obtain the document reference and update the catalog baa_valid field to true. If no BAA exists, initiate BAA negotiation with Volpara Health. This is the single highest-leverage
+15% confidence -0% ROI variance 👤 📋 · 3 pts
APP-034-002 Obtain Volpara Health Cloud SaaS contract intelligence, pricing, and migration terms
P1 Procurement
Engage Volpara Health to obtain: (1) a customer-specific SaaS migration proposal or RFP response for Volpara Health Cloud, (2) annual contract value (ACV) and subscription pricing for the SaaS product, (3) data portability and migration support terms, (4) contract renewal date for the current on-premises virtual appliance license, and (5) HITRUST CSF certification status for Volpara Health Cloud.
+0% confidence -0% ROI variance 👤 📋 · 5 pts
APP-034-003 Formally map DICOM, PACS/VNA, and RIS integration dependencies
P2 Dependency Mapping
Conduct a formal integration mapping exercise to confirm and document all upstream and downstream dependencies for Volpara: (1) DICOM modality feed — identify the mammography modality or PACS/VNA system sending DICOM images to Volpara, including protocol version, port, and network path; (2) PACS/VNA — confirm the picture archiving system providing image retrieval and the integration method (DICOM
+0% confidence -0% ROI variance 👤 📋 · 5 pts
APP-034-004 Collect DR and RTO/RPO requirements for Volpara against SaaS SLA
P2 Architecture
Obtain the formal DR requirements (RTO and RPO targets) for Volpara from the application owner or IT risk team. This is a High-criticality, PHI-handling workload with no DR data on file, resulting in the maximum DR penalty (10%) applied to ROI variance. Once RTO/RPO targets are known, validate them against the Volpara Health Cloud SaaS SLA (uptime guarantee, backup frequency, geo-redundancy, and i
+0% confidence -0% ROI variance 👤 📋 · 3 pts
APP-034-005 Deploy 30-day APM telemetry baseline on volparaprd VM
P2 Telemetry
Install or enable APM/infrastructure monitoring on the volparaprd VM (Ubuntu Linux 64-bit) to collect a minimum 30-day time-series of CPU utilization, memory active percentage, disk I/O, and network throughput. This resolves the SRC-CAP-004 constraint (snapshot_count=1, apm_telemetry_available=false) which currently caps right-sizing at MODERATE (SRC-RS-002, 25% reduction) and prevents ZOMBIE or O
+0% confidence -0% ROI variance 👤 📋 · 3 pts
APP-034-006 Correct OS catalog discrepancy: Windows Server vs. Ubuntu Linux
P3 Architecture
The application catalog records OS as 'Windows Server' for app-034 (Volpara), but the RVTools export for VM 'volparaprd' reports Ubuntu Linux (64-bit). This discrepancy affects patching strategy, migration agent selection, APM agent deployment, and security compliance classification. Verify the actual OS by logging into the VM or reviewing the hypervisor console. Update the application catalog to
+0% confidence -0% ROI variance 👤 📋 · 1 pts
This tab shows the evidence behind the recommendation. The seven dimensions are scored by quality — CONFIRMED evidence counts fully, INFERRED evidence counts less. The validator table confirms each analysis step passed its checks. Green means the step is sound.
Evidence Classification — 7 Dimensions
Dimension Classification Multiplier Contribution
Telemetry SINGLE 0.50 6.00
Dependency Mapping INFERRED 0.25 2.50
Procurement Contracts MISSING 0.00 0.00
Architecture Diagrams INFERRED 0.25 1.75
Security Compliance INFERRED 0.25 1.25
Business Criticality SINGLE 0.50 2.00
Dr Requirements MISSING 0.00 0.00
Applied Cap: SRC-CAP-001
Validation Reports — Per Agent
Agent Status Provenance Counts Defaults Used Rule Failures
telemetry PASS
SRC:2CUST:9COMP:3DEF:2
 None ✓ None
dependency PASS
SRC:8COMP:5DEF:2
 None ✓ None
procurement PASS
None ✓ None
provisioning PASS_WITH_DEFAULTS
SRC:7CUST:1COMP:9DEF:1
• {'default_id': 'DEF-MIG-001', 'field': 'migration_factor', 'value': 0.2, 'reason
 ✓ None
synthesizer PASS
None ✓ None
confidence_advisor PASS
COMP:18
 None ✓ None
This section contains the full technical reasoning behind the recommendation: the rules that fired, the decision path, the confidence math, and the open intelligence gaps. It is intended for analysts and engineers. Executive readers can rely on the Overview tab.
Rules Applied
These are the specific governance rules that produced this recommendation.
▸ RULE 1 — REPURCHASE
A vendor SaaS replacement is available and credible, and the app has no hardware constraint — so the recommendation is to replace it with the vendor's cloud service.
▸ SRC-CAP-001 (binding)
Confidence is capped at 40 because the app handles PHI and the vendor's Business Associate Agreement is not confirmed. PHI cannot move to a cloud vendor without a confirmed BAA.
▸ SRC-CAP-002
Confidence is capped at 55 because more than half of the seven evidence dimensions are inferred or missing — the conclusion is data-thin.
▸ SRC-CAP-003
Confidence is capped at 65 because none of the seven evidence dimensions reached CONFIRMED quality — there is no high-quality anchor for the analysis.
▸ SRC-CAP-004
Confidence is capped at 65 because telemetry is a single point-in-time snapshot with no ongoing monitoring — right-sizing needs usage patterns over time.
▸ RULE T2 — Right-Sizing Hard Cap
Right-sizing is limited to a moderate 25% reduction because telemetry is a single snapshot. More aggressive right-sizing needs 30+ days of monitoring data.
Decision Path
The step-by-step trace of how the recommendation was reached, in the system's own terms.
PRE-CHECK: hardware_dependency=false, life_safety_classification=false — structural blocker does not fire. RULE 1 (REPURCHASE): saas_alternative_available=true (Volpara Health Cloud per notes + procurement), saas_confidence_inferred=72 >= 50, hardware_dependency != true — RULE 1 FIRES. strategic_recommendation=REPURCHASE. CONFIDENCE: base=50, telemetry(SINGLE)=+6.0, dependency_mapping(INFERRED)=+2.5, procurement_contracts(MISSING)=+0.0, architecture_diagrams(INFERRED)=+1.75, security_compliance(INFERRED)=+1.25, business_criticality(SINGLE)=+2.0, dr_requirements(MISSING)=+0.0 → pre_cap=63.5. CAP EVALUATION: SRC-CAP-001 (phi=true, baa!=true)→40; SRC-CAP-002 (5/7 INFERRED/MISSING)→55; SRC-CAP-003 (0 CONFIRMED)→65; SRC-CAP-004 (snapshot=1, apm=false)→65. Lowest cap: SRC-CAP-001=40. final_confidence=40. GATE (v4.1): final_confidence=40 < 60, strategic_recommendation=REPURCHASE (not RETAIN) → recommendation_status=PENDING_CONDITIONS, confidence_gate_applied=true, displayed_primary=REPURCHASE (gate does NOT force RETAIN). Conditions to unlock: (1) Confirm BAA with Volpara Health for Volpara Health Cloud; (2) Obtain contract intelligence and SaaS pricing; (3) Map DICOM/PACS/RIS integration dependencies formally.
Cap Evaluation (all fired caps): SRC-CAP-001: PHI handled + BAA not confirmed → ceiling 40 (ceiling=40) ← BINDING SRC-CAP-002: ≥4 of 7 dimensions INFERRED/MISSING → ceiling 55 (ceiling=55) SRC-CAP-003: Zero CONFIRMED dimensions → ceiling 65 (ceiling=65) SRC-CAP-004: Single snapshot, no APM → ceiling 65 (ceiling=65)
Confidence Scoring Breakdown
How the confidence percentage was calculated, dimension by dimension.
Base score: 50 Evidence contributions: Telemetry SINGLE +6.00 Dependency Mapping INFERRED +2.50 Procurement Contracts MISSING +0.00 Architecture Diagrams INFERRED +1.75 Security Compliance INFERRED +1.25 Business Criticality SINGLE +2.00 Dr Requirements MISSING +0.00 Pre-cap score: 63.5 Binding cap applied: SRC-CAP-001 Final confidence: 40%
Intelligence Gaps
Specific things we don't yet know. Each gap is a reason the confidence score is held back.
BAA Confirmation for Volpara Health Cloud
phi_handled=true with baa_confirmed=UNKNOWN triggers SRC-CAP-001, capping confidence at 40 and blocking EXECUTABLE status. Must confirm BAA exists for vendor SaaS product before any migration action.
Contract Intelligence — SaaS Licensing and Migration Terms
repurchase_viable=INSUFFICIENT_EVIDENCE; no contract data available. SaaS pricing, migration support terms, and data portability clauses unknown. Procurement penalty of 15% applied to ROI variance.
DICOM / PACS / RIS Integration Mapping
All three integration dependencies (DICOM modality feed, PACS/VNA, RIS) are INFERRED from vendor product knowledge. Formal integration mapping required to validate cutover sequencing and confirm no hidden hardware or latency constraints.
DR and RTO/RPO Requirements
No DR data provided for a High-criticality PHI-handling workload. DR penalty applied at maximum (10%). SaaS SLA must be validated against MUSC DR requirements.
APM Telemetry Baseline
apm_telemetry_available=false; right-sizing confidence limited to single RVTools snapshot. SRC-CAP-004 applies. Performance baseline for SaaS sizing cannot be confirmed without application-level telemetry.
Technical Detail
A fuller technical explanation of the recommendation and its constraints.
Procurement confirms Volpara Health Cloud as a vendor-hosted SaaS alternative with saas_confidence_inferred=72, satisfying Rule 1 threshold of 50. Dependency analysis confirms hardware_dependency=false and life_safety_classification=false, clearing the structural pre-check. The application is a single-VM virtual appliance (volparaprd) running mammography AI analytics — a workload the vendor has explicitly productized as SaaS. SRC-CAP-001 applies because phi_handled=true and baa_confirmed=UNKNOWN, capping confidence at 40 and placing the recommendation in PENDING_CONDITIONS status until BAA and contract terms are confirmed.
Secondary Path
The fallback option if the primary recommendation cannot proceed.
Secondary Path — RETAIN
RETAIN is the secondary disposition if BAA cannot be confirmed for Volpara Health Cloud and no alternative PHI-compliant SaaS path is identified; the existing virtual appliance would remain on-premises with operational monitoring improvements.
Right-Sizing Notes
Why the cloud sizing estimate is set where it is, based on the telemetry available.
Hard rule T2 triggered: snapshot_count=1 AND apm_telemetry_available=false. ZOMBIE (SRC-RS-004) and OVERSIZED (SRC-RS-003) tiers are forbidden regardless of utilization metrics. Fleet avg mem_active_pct is 35.0%, which under Rule T3 would independently yield MODERATE (SRC-RS-002, 0.25 reduction); the hard-rule cap and utilization-based tier are therefore consistent. CPU utilization data is unavailable (cpu_avg_pct null on 1 of 1 VMs). Rule T1 override does not apply: application_type is Virtual Appliance, not Hardware-Integrated, and life_safety_classification is absent from the catalog. Right-sizing opportunity exists at MODERATE tier (0.25 reduction) pending additional telemetry to confirm.
This tab shows the raw output of each analysis agent — the specialized components that examine telemetry, dependencies, contracts, and cost. It is the most detailed and most technical view. Most readers will find the Overview and Backlog tabs sufficient.
Telemetry Agent
Utilization Profilemoderate
Snapshot Count1
APM AvailableFalse
Hard Rule TriggeredTrue
Right-Sizing Opp.True
Agent Confidence50%
Telemetry — VM Counts
Powered Off VMs0
Flagged VMs0
Mem Active % (avg)35.0%
Data WindowNone
Inventory Source
Dependency Agent
Epic IntegrationFalse
PHI HandledTrue
Hardware DependencyFalse
Latency SensitiveFalse
Blast Radiusmedium
Migration RiskMEDIUM
Conf (Inferred)72%
Conf (Confirmed)55%
Verificationsingle_source
Cloud Connectivity
ComplianceHIPAA / HITRUST
Upstream (2)
DICOM / Mammography Imaging Modality Feed: inferred: Volpara product function (mammography AI analytics requires DICOM image input from mammography modalities or…
PACS / VNA (Picture Archiving and Communication System): inferred: standard mammography analytics integration pattern; Volpara typically integrates with PACS for image retrieval
Downstream (1)
Radiology Reporting / RIS: inferred: Volpara density scores and risk flags are typically consumed by radiology information systems or EMR for…
Procurement Agent
SaaS Available—
SaaS VendorVolpara Health
SaaS ProductVolpara Health Cloud
Repurchase ViableINSUFFICIENT_EVIDENCE
BAA ConfirmedUNKNOWN
SaaS Confidence
Inferred Score72%
Confirmed Score0%
Verificationsingle_source
AWS Equiv.None
Provisioning Agent (Python engine — not LLM)
Current VMs1
Recommended VMs1
Right-Sizing25% reduction
Current TCO/yr$6,500
Projected Cloud TCO$5,200
Migration One-Time$1,040
Dual-Running Cost$3,900
-$3,640
Year-1 Net
$1,300
Steady-State
-$1,040
3-Year
AWS TargetAmazon EC2
Azure TargetAzure Virtual Machines
Cost Modeldefault_multipliers
DEF: {'default_id': 'DEF-MIG-001', 'field': 'migration_factor', 'value': 0.2, 'reason': 'migration_complexity not provided'}
DEF: {'default_id': 'DEF-COST-001', 'field': 'cost_model_basis', 'value': 'default_multipliers', 'reason': 'no customer cloud quote provided; using SRC-TCO-* and SRC-MULT-* registry defaults'}
ACRConnect
Vendor: ACR · 1 VMs · Criticality: High · HIPAA / HITRUST
REPURCHASE
55%
Confidence
28%
Data Quality
ROI: Year-1 Net: -$3,640 · Steady-State: $1,300/yr · 3-Year: -$1,040
Overview
Backlog
Provenance
Technical Details
Agents
Recommendation
REPURCHASE ⏳ PENDING CONDITIONS
SRC-CAP-002 Binding cap · Ceiling: 55% · Threshold: 60%
Secondary: RETAIN
Confidence
55%
Viable, but specific gaps must close first.
ROI Summary (Python-computed, not LLM)
-$3,640
Year-1 Net
$1,300
Steady-State /yr
-$1,040
3-Year Total
ROI variance: ±55%
⚙ Financials computed by provisioning_engine_v3 (Python) — not LLM-generated
In plain terms: migrating costs money up front — Year 1 is a net outflow of $3,640. From Year 2 onward the app saves $1,300 every year. The 3-Year total nets these against each other.
Finding
ACRConnect has a direct same-vendor SaaS successor — ACRConnect Cloud — making this a strong REPURCHASE candidate that would eliminate the on-premises VM footprint entirely. However, the PHI handling status and Business Associate Agreement with ACR have not been confirmed, which is a prerequisite for any cloud migration in a HIPAA/HITRUST environment. Completing BAA verification and an integration inventory for the RIS/PACS/HL7 feeds will unlock this migration.
In plain terms: this app should move to the vendor's cloud product, but a few specific items must be confirmed first. See Conditions to Unlock below.
Conditions to Unlock (3 items)
This recommendation becomes executable once these conditions are confirmed:
☐ Verify whether ACRConnect stores or transmits patient health information
Owner: Compliance Officer / Application Owner (Erika Murphy) · Timeline: 7 days · Unlocks: compliance_close, cap_resolution
☐ Confirm a signed data privacy agreement with ACR for the cloud product
Owner: Legal / Compliance Officer · Timeline: 21 days · Unlocks: compliance_close, cap_resolution, migration_commit
☐ Get pricing and contract details for ACRConnect Cloud from the vendor
Owner: Procurement / Vendor Manager · Timeline: 21 days · Unlocks: cost_clarity, migration_commit, cap_resolution
See Backlog tab for full action backlog and P2/P3 items.
This is the list of work needed to move this recommendation forward. The 3 P1 items are the priority — they unlock the recommendation. P2 and P3 items improve confidence and reduce risk but are not blockers. 'Story points' measure effort, not time or cost — see the Methodology tab for how they work.
55%
Current
60%
Threshold
62%
Ceiling
17
Story Points
Why This Score
ACRConnect is capped at 55 by SRC-CAP-002 because 5 of 7 evidence dimensions (dependency_mapping, procurement_contracts, architecture_diagrams, security_compliance, dr_requirements) are INFERRED or MISSING, leaving the pre-cap score of 61.75 suppressed to the majority-inferred ceiling. With zero available headroom (current_confidence equals confidence_ceiling), no backlog item can contribute a confidence delta in this run; all items target cap resolution and variance reduction to unlock headroom in the next synthesizer cycle.
Proceeding with REPURCHASE to ACRConnect Cloud without resolving BAA confirmation and PHI classification exposes the organization to HIPAA violation risk and potential OCR enforcement action, as PHI may be transmitted to a SaaS environment without a legally required Business Associate Agreement.
Action Backlog — 6 Items · 17 Story Points · Ceiling: 62%
APP-035-001 Confirm PHI Handling Classification via Compliance Officer Attestation
P1 Security/Compliance
The catalog field phi_handling=true uses a non-canonical field name and cannot satisfy the v3.1 tri-state requirement for phi_handled. A compliance officer or application owner (POC: Erika Murphy) must provide a written attestation confirming whether ACRConnect stores, processes, or transmits PHI (patient demographics, imaging orders, DICOM metadata, or HL7 results). This converts phi_handled from
+7% confidence -0% ROI variance 👤 📋 · 1 pts
APP-035-002 Obtain Executed BAA with ACR Covering ACRConnect Cloud
P1 BAA/Legal
baa_confirmed=UNKNOWN because application_catalog.baa_valid is null. In a HIPAA/HITRUST environment with likely PHI handling, a signed Business Associate Agreement with ACR covering ACRConnect Cloud is a legal prerequisite before any PHI can be processed in the SaaS environment. Legal/Compliance must (1) determine whether a current BAA exists for the on-premises product, (2) confirm whether it ext
+0% confidence -0% ROI variance 👤 📋 · 2 pts
APP-035-003 Engage ACR Procurement for SaaS Contract Terms and ACV
P1 Procurement
procurement_contracts=MISSING; contract_acv=null; contract_renewal_date=null. The ROI model carries a 15-point procurement penalty and the three-year savings projection is negative (-$1,040) under default assumptions. Engaging ACR to obtain a formal SaaS migration proposal, ACV, renewal date, and migration pricing will (1) replace the default cloud multiplier (0.8) with an actual SaaS subscription
+7% confidence -0% ROI variance 👤 📋 · 2 pts
APP-035-004 Conduct Integration Inventory: RIS, PACS, and HL7/DICOM Feed Mapping
P2 Dependency Mapping
dependency_mapping=INFERRED; upstream dependencies (RIS/PACS, HL7/DICOM ADT feeds) are inferred from vendor and application name patterns only. A confirmed integration map is required to (1) assess cutover complexity for ACRConnect Cloud migration, (2) validate ACRConnect Cloud compatibility with existing imaging infrastructure, (3) confirm or refute the blast_radius=high assessment, and (4) ident
+0% confidence -0% ROI variance 👤 📋 · 5 pts
APP-035-005 Deploy 30-Day APM Telemetry Collection for Workload Characterization
P2 Telemetry
apm_telemetry_available=false; snapshot_count=1; SRC-CAP-004 is active (ceiling 65). The current right-sizing tier is capped at MODERATE (SRC-RS-002, 25% reduction) per hard rule T2 because a single RVTools snapshot cannot distinguish a genuinely idle VM from one sampled between bursts. CPU utilization data is entirely absent. Deploying a lightweight APM agent (e.g., CloudWatch Agent, Azure Monito
+0% confidence -0% ROI variance 👤 📋 · 5 pts
APP-035-006 Document DR Requirements and RTO/RPO for ACRConnect
P3 Architecture
dr_requirements=MISSING and architecture_diagrams=MISSING; both carry zero contribution to confidence scoring and add a combined 25-point penalty to roi_variance_pct (15 for architecture MISSING + 10 for DR MISSING). For a High-criticality radiology connectivity platform in a HIPAA/HITRUST environment, DR requirements (RTO, RPO, backup frequency, failover architecture) must be documented to (1) va
+7% confidence -0% ROI variance 👤 📋 · 2 pts
This tab shows the evidence behind the recommendation. The seven dimensions are scored by quality — CONFIRMED evidence counts fully, INFERRED evidence counts less. The validator table confirms each analysis step passed its checks. Green means the step is sound.
Evidence Classification — 7 Dimensions
Dimension Classification Multiplier Contribution
Telemetry SINGLE 0.50 6.00
Dependency Mapping INFERRED 0.25 2.50
Procurement Contracts MISSING 0.00 0.00
Architecture Diagrams MISSING 0.00 0.00
Security Compliance INFERRED 0.25 1.25
Business Criticality SINGLE 0.50 2.00
Dr Requirements MISSING 0.00 0.00
Applied Cap: SRC-CAP-002
Validation Reports — Per Agent
Agent Status Provenance Counts Defaults Used Rule Failures
telemetry PASS
SRC:2CUST:10COMP:4DEF:2
 None ✓ None
dependency PASS
SRC:1CUST:5COMP:2DEF:6
 None ✓ None
procurement PASS
None ✓ None
provisioning PASS_WITH_DEFAULTS
SRC:7CUST:1COMP:9DEF:1
• {'default_id': 'DEF-WL-001', 'field': 'workload_type', 'value': 'app_server', 'r
• {'default_id': 'DEF-MIG-001', 'field': 'migration_factor', 'value': 0.2, 'reason
 ✓ None
synthesizer PASS
None ✓ None
confidence_advisor PASS
COMP:33
 None ✓ None
This section contains the full technical reasoning behind the recommendation: the rules that fired, the decision path, the confidence math, and the open intelligence gaps. It is intended for analysts and engineers. Executive readers can rely on the Overview tab.
Rules Applied
These are the specific governance rules that produced this recommendation.
▸ RULE 1 — REPURCHASE
A vendor SaaS replacement is available and credible, and the app has no hardware constraint — so the recommendation is to replace it with the vendor's cloud service.
▸ SRC-CAP-002 (binding)
Confidence is capped at 55 because more than half of the seven evidence dimensions are inferred or missing — the conclusion is data-thin.
▸ SRC-CAP-003
Confidence is capped at 65 because none of the seven evidence dimensions reached CONFIRMED quality — there is no high-quality anchor for the analysis.
▸ SRC-CAP-004
Confidence is capped at 65 because telemetry is a single point-in-time snapshot with no ongoing monitoring — right-sizing needs usage patterns over time.
▸ RULE T2 — Right-Sizing Hard Cap
Right-sizing is limited to a moderate 25% reduction because telemetry is a single snapshot. More aggressive right-sizing needs 30+ days of monitoring data.
Decision Path
The step-by-step trace of how the recommendation was reached, in the system's own terms.
PRE-CHECK: hardware_dependency=false → structural blocker does not fire. RULE 1 (REPURCHASE): saas_available=true (CUSTOMER_PROVIDED via catalog notes + procurement agent), saas_confidence_inferred=72 ≥ 50 threshold, hardware_dependency=false → RULE 1 FIRES. strategic_recommendation=REPURCHASE. CONFIDENCE SCORING: base=50; telemetry=SINGLE(+6.00), dependency_mapping=INFERRED(+2.50), procurement_contracts=MISSING(+0), architecture_diagrams=MISSING(+0), security_compliance=INFERRED(+1.25), business_criticality=SINGLE(+2.00), dr_requirements=MISSING(+0); pre_cap=61.75. CAP EVALUATION: SRC-CAP-001 not triggered (phi_handled≠confirmed true); SRC-CAP-002 triggered (5/7 dimensions INFERRED/MISSING → cap 55); SRC-CAP-003 triggered (zero CONFIRMED → cap 65); SRC-CAP-004 triggered (snapshot_count=1, apm=false → cap 65); binding cap = SRC-CAP-002 at 55. final_confidence=55. GATE (v4.1): final_confidence=55 < 60 AND strategic_recommendation=REPURCHASE (not RETAIN) → recommendation_status=PENDING_CONDITIONS; displayed_primary=REPURCHASE (gate does NOT force RETAIN in v4.1); confidence_gate_applied=true.
Cap Evaluation (all fired caps): SRC-CAP-002: ≥4 of 7 dimensions INFERRED/MISSING → ceiling 55 (ceiling=55) ← BINDING SRC-CAP-003: Zero CONFIRMED dimensions → ceiling 65 (ceiling=65) SRC-CAP-004: Single snapshot, no APM → ceiling 65 (ceiling=65)
Confidence Scoring Breakdown
How the confidence percentage was calculated, dimension by dimension.
Base score: 50 Evidence contributions: Telemetry SINGLE +6.00 Dependency Mapping INFERRED +2.50 Procurement Contracts MISSING +0.00 Architecture Diagrams MISSING +0.00 Security Compliance INFERRED +1.25 Business Criticality SINGLE +2.00 Dr Requirements MISSING +0.00 Pre-cap score: 61.75 Binding cap applied: SRC-CAP-002 Final confidence: 55%
Intelligence Gaps
Specific things we don't yet know. Each gap is a reason the confidence score is held back.
BAA Confirmation with ACR (SaaS vendor)
baa_confirmed=UNKNOWN in a HIPAA/HITRUST environment; a signed BAA with ACR covering ACRConnect Cloud is a legal prerequisite before any PHI can be processed in the SaaS environment. Without this, REPURCHASE cannot proceed.
PHI Handling Classification
phi_handled=UNKNOWN despite ACRConnect being a radiology connectivity platform in a healthcare environment. Radiology workflows routinely involve PHI (patient demographics, imaging orders, results). Classification must be confirmed to determine compliance obligations for SaaS migration.
Integration Inventory: RIS / PACS / HL7 Feeds
Upstream dependencies (RIS/PACS, HL7/DICOM ADT feeds) are inferred from vendor and application name patterns only. A confirmed integration map is required to assess cutover complexity and validate ACRConnect Cloud compatibility with existing imaging infrastructure.
Contract Intelligence and Renewal Timeline
contract_intelligence=INSUFFICIENT_EVIDENCE; no data on current license term, renewal date, or SaaS migration pricing. Contract leverage window is unknown, which affects migration scheduling and cost negotiation.
APM Telemetry Availability
apm_telemetry_available=false; right-sizing is based on a single RVTools snapshot showing 1% average memory utilization on a 32 GiB / 8 vCPU VM. Right-sizing tier is capped at MODERATE (SRC-RS-002) per hard rule. Actual workload characterization requires application-level telemetry.
Technical Detail
A fuller technical explanation of the recommendation and its constraints.
Rule 1 (REPURCHASE) fires: saas_alternative_available=true (confirmed via application_catalog.notes and procurement agent), saas_confidence_inferred=72 ≥ 50, and hardware_dependency=false. The pre-check structural blocker does not apply. The procurement agent confirms ACRConnect Cloud as the same-vendor SaaS product with inferred confidence of 72, satisfying the v4.1 lowered threshold. Confidence is capped at 55 by SRC-CAP-002 (5 of 7 dimensions INFERRED or MISSING), triggering PENDING_CONDITIONS status rather than EXECUTABLE. The secondary disposition is RETAIN, reflecting the structural fallback if BAA and PHI conditions cannot be resolved.
Secondary Path
The fallback option if the primary recommendation cannot proceed.
Secondary Path — RETAIN
RETAIN is the secondary disposition: hardware_dependency=false and no life-safety classification means there is no structural barrier, but unresolved BAA and PHI handling status create compliance risk that could prevent SaaS migration, leaving on-premises operation as the interim state.
Right-Sizing Notes
Why the cloud sizing estimate is set where it is, based on the telemetry available.
Raw mem_active_pct of 1.0% would ordinarily trigger SRC-RS-004 (ZOMBIE) under RULE T3. However, RULE T2 hard rule fires unconditionally: snapshot_count=1 AND apm_telemetry_available=false. Single-snapshot data cannot distinguish a genuinely idle VM from one sampled between bursts. Tier is capped at SRC-RS-002 (MODERATE, 0.25 reduction). SRC-RS-003 and SRC-RS-004 are FORBIDDEN for this run. CPU utilization data is entirely absent (cpu_avg_pct null on 1 of 1 VMs); no APM time-series available. Right-sizing opportunity is flagged at MODERATE confidence only. Telemetry depth is insufficient to support a stronger recommendation.
This tab shows the raw output of each analysis agent — the specialized components that examine telemetry, dependencies, contracts, and cost. It is the most detailed and most technical view. Most readers will find the Overview and Backlog tabs sufficient.
Telemetry Agent
Utilization Profilemoderate
Snapshot Count1
APM AvailableFalse
Hard Rule TriggeredTrue
Right-Sizing Opp.True
Agent Confidence42%
Telemetry — VM Counts
Powered Off VMs0
Flagged VMs0
Mem Active % (avg)1.0%
Data WindowNone
Inventory Source
Dependency Agent
Epic IntegrationFalse
PHI HandledUNKNOWN
Hardware DependencyFalse
Latency SensitiveUNKNOWN
Blast Radiushigh
Migration RiskMEDIUM
Conf (Inferred)72%
Conf (Confirmed)38%
Verificationsingle_source
Cloud Connectivity
ComplianceHIPAA / HITRUST
Upstream (2)
Radiology Information System (RIS) / PACS: inferred: ACRConnect (ACR vendor) is a radiology connectivity platform; structural pattern implies upstream imaging…
HL7 / DICOM feed (likely Epic or standalone ADT): inferred: radiology connectivity platforms in healthcare typically receive HL7 ADT or DICOM worklist feeds;…
Downstream (2)
ACRConnect Cloud (SaaS alternative candidate): confirmed: application_catalog.notes = 'SaaS alternative: ACRConnect Cloud.'
Downstream imaging consumers (PACS viewers, reporting systems): inferred: radiology connectivity platform structural pattern implies downstream consumers; none explicitly named in…
Procurement Agent
SaaS Available—
SaaS VendorACR
SaaS ProductACRConnect Cloud
Repurchase ViableINSUFFICIENT_EVIDENCE
BAA ConfirmedUNKNOWN
SaaS Confidence
Inferred Score72%
Confirmed Score0%
Verificationsingle_source
AWS Equiv.None
Provisioning Agent (Python engine — not LLM)
Current VMs1
Recommended VMs1
Right-Sizing25% reduction
Current TCO/yr$6,500
Projected Cloud TCO$5,200
Migration One-Time$1,040
Dual-Running Cost$3,900
-$3,640
Year-1 Net
$1,300
Steady-State
-$1,040
3-Year
AWS TargetAmazon EC2
Azure TargetAzure Virtual Machines
Cost Modeldefault_multipliers
DEF: {'default_id': 'DEF-WL-001', 'field': 'workload_type', 'value': 'app_server', 'reason': 'workload_type indeterminable'}
DEF: {'default_id': 'DEF-MIG-001', 'field': 'migration_factor', 'value': 0.2, 'reason': 'migration_complexity not provided'}
DEF: {'default_id': 'DEF-COST-001', 'field': 'cost_model_basis', 'value': 'default_multipliers', 'reason': 'no customer cloud quote provided; using SRC-TCO-* and SRC-MULT-* registry defaults'}
Teamplay
Vendor: Siemens Healthineers · 1 VMs · Criticality: High · HIPAA / HITRUST
REPURCHASE
55%
Confidence
32%
Data Quality
ROI: Year-1 Net: -$3,640 · Steady-State: $1,300/yr · 3-Year: -$1,040
Overview
Backlog
Provenance
Technical Details
Agents
Recommendation
REPURCHASE ⏳ PENDING CONDITIONS
SRC-CAP-002 Binding cap · Ceiling: 55% · Threshold: 60%
Secondary: RETAIN
Confidence
55%
Viable, but specific gaps must close first.
ROI Summary (Python-computed, not LLM)
-$3,640
Year-1 Net
$1,300
Steady-State /yr
-$1,040
3-Year Total
ROI variance: ±60%
⚙ Financials computed by provisioning_engine_v3 (Python) — not LLM-generated
In plain terms: migrating costs money up front — Year 1 is a net outflow of $3,640. From Year 2 onward the app saves $1,300 every year. The 3-Year total nets these against each other.
Finding
Siemens Teamplay has a direct vendor-hosted SaaS successor — Siemens Teamplay Cloud — which would eliminate the single on-premises Windows Server VM and keep the application under the same vendor support umbrella. The transition is strategically sound but cannot be confirmed as executable today because the Business Associate Agreement status and contract terms are unresolved, and PHI handling through the SaaS platform has not been verified. Completing BAA confirmation and a PHI data-flow review are the two conditions that would unlock this migration.
In plain terms: this app should move to the vendor's cloud product, but a few specific items must be confirmed first. See Conditions to Unlock below.
Conditions to Unlock (3 items)
This recommendation becomes executable once these conditions are confirmed:
☐ Verify Business Associate Agreement for Siemens Teamplay Cloud
Owner: Compliance Officer / Legal · Timeline: 10 days · Unlocks: compliance_close, cap_resolution, migration_commit
☐ Map and verify PHI data flows for Teamplay on-premises and cloud platforms
Owner: Security / HIPAA Privacy Officer · Timeline: 15 days · Unlocks: compliance_close, cap_resolution, migration_commit
☐ Document Teamplay system architecture and integration points
Owner: Enterprise Architect / Application Owner · Timeline: 10 days · Unlocks: cap_resolution, migration_commit, operational_safety
See Backlog tab for full action backlog and P2/P3 items.
This is the list of work needed to move this recommendation forward. The 3 P1 items are the priority — they unlock the recommendation. P2 and P3 items improve confidence and reduce risk but are not blockers. 'Story points' measure effort, not time or cost — see the Methodology tab for how they work.
55%
Current
60%
Threshold
64%
Ceiling
22
Story Points
Why This Score
Confidence is capped at 55 by SRC-CAP-002 because 5 of 7 evidence dimensions (dependency_mapping, procurement_contracts, architecture_diagrams, security_compliance, dr_requirements) are INFERRED or MISSING, with zero CONFIRMED dimensions triggering SRC-CAP-003 as well. The pre-cap score of 63.75 would clear the 60-point threshold, meaning the recommendation is structurally sound but cannot be acted upon until the evidence base is upgraded from inferred to confirmed.
Proceeding with REPURCHASE to Siemens Teamplay Cloud without resolving BAA confirmation and PHI data-flow verification exposes MUSC to a material HIPAA violation risk, as PHI transmission to a vendor-hosted platform without an executed BAA is non-compliant regardless of the strategic merit of the SaaS transition.
Action Backlog — 6 Items · 22 Story Points · Ceiling: 64%
APP-036-001 Confirm BAA status with Siemens Healthineers for Teamplay Cloud
P1 BAA/Legal
Engage MUSC Compliance Officer and Siemens Healthineers account team to confirm whether a valid BAA exists or can be executed covering the vendor-hosted Teamplay Cloud environment. Obtain BAA document or written confirmation of BAA status. If no BAA exists, initiate BAA negotiation with Siemens Healthineers. Resolution is a hard prerequisite for REPURCHASE execution under HIPAA — without baa_confi
+0% confidence -0% ROI variance 👤 📋 · 3 pts
APP-036-002 Conduct formal PHI data-flow assessment for Teamplay and Teamplay Cloud
P1 Security/Compliance
Commission a formal PHI data-flow assessment to: (1) confirm which PHI categories transit through Siemens Teamplay (imaging metadata, patient identifiers, study data); (2) verify that Siemens Teamplay Cloud's data residency, encryption, and processing controls satisfy MUSC's HIPAA/HITRUST obligations; (3) assess whether de-identification is feasible for any data streams; (4) resolve phi_handled tr
+9% confidence -0% ROI variance 👤 📋 · 5 pts
APP-036-003 Capture architecture diagrams and integration topology for Teamplay
P1 Architecture
Produce current-state architecture diagrams for Siemens Teamplay covering: (1) network topology and firewall zones; (2) upstream PACS/modality worklist integration pattern (DICOM push vs. pull, HL7 feed configuration); (3) Active Directory / Windows authentication integration; (4) downstream analytics consumer connections; (5) data flows including PHI-bearing streams. Architecture diagrams are cur
+9% confidence -0% ROI variance 👤 📋 · 5 pts
APP-036-004 Obtain Siemens Teamplay contract terms and Teamplay Cloud migration pricing
P2 Procurement
Engage Siemens Healthineers account team to obtain: (1) current on-premises Teamplay contract ACV and renewal date; (2) Teamplay Cloud SaaS subscription pricing and migration proposal; (3) any BAA novation or amendment required for the cloud-hosted environment (coordinate with APP-036-001). Contract intelligence is currently INFERRED — resolving to CONFIRMED will promote the procurement_contracts
+9% confidence -0% ROI variance 👤 📋 · 3 pts
APP-036-005 Confirm upstream PACS and modality worklist integration dependencies
P2 Dependency Mapping
Confirm the upstream PACS/modality worklist integration that is currently inferred from vendor product knowledge. Specifically: (1) identify the PACS system(s) feeding Teamplay (vendor, version, integration protocol); (2) confirm DICOM push vs. pull configuration and any HL7 feed details; (3) assess whether Teamplay Cloud requires a different integration pattern that would increase migration compl
+0% confidence -0% ROI variance 👤 📋 · 3 pts
APP-036-006 Define DR, RTO, and RPO requirements for Teamplay
P2 Architecture
Engage application owner (Wendy Ketchum) and clinical operations stakeholders to define and document DR requirements for Siemens Teamplay: (1) Recovery Time Objective (RTO) and Recovery Point Objective (RPO); (2) whether Teamplay Cloud's published SLA meets MUSC's operational continuity standards for a High-criticality clinical analytics platform; (3) any regulatory DR requirements under HIPAA/HIT
+9% confidence -0% ROI variance 👤 📋 · 3 pts
This tab shows the evidence behind the recommendation. The seven dimensions are scored by quality — CONFIRMED evidence counts fully, INFERRED evidence counts less. The validator table confirms each analysis step passed its checks. Green means the step is sound.
Evidence Classification — 7 Dimensions
Dimension Classification Multiplier Contribution
Telemetry SINGLE 0.50 6.00
Dependency Mapping INFERRED 0.25 2.50
Procurement Contracts INFERRED 0.25 2.00
Architecture Diagrams MISSING 0.00 0.00
Security Compliance INFERRED 0.25 1.25
Business Criticality SINGLE 0.50 2.00
Dr Requirements MISSING 0.00 0.00
Applied Cap: SRC-CAP-002
Validation Reports — Per Agent
Agent Status Provenance Counts Defaults Used Rule Failures
telemetry PASS
SRC:2CUST:9COMP:3DEF:2
 None ✓ None
dependency PASS
SRC:1CUST:3COMP:3DEF:3
 None ✓ None
procurement PASS
None ✓ None
provisioning PASS_WITH_DEFAULTS
SRC:7CUST:1COMP:9DEF:1
• {'default_id': 'DEF-WL-001', 'field': 'workload_type', 'value': 'app_server', 'r
• {'default_id': 'DEF-MIG-001', 'field': 'migration_factor', 'value': 0.2, 'reason
 ✓ None
synthesizer PASS
None ✓ None
confidence_advisor PASS
COMP:15
 None ✓ None
This section contains the full technical reasoning behind the recommendation: the rules that fired, the decision path, the confidence math, and the open intelligence gaps. It is intended for analysts and engineers. Executive readers can rely on the Overview tab.
Rules Applied
These are the specific governance rules that produced this recommendation.
▸ RULE 1 — REPURCHASE
A vendor SaaS replacement is available and credible, and the app has no hardware constraint — so the recommendation is to replace it with the vendor's cloud service.
▸ SRC-CAP-002 (binding)
Confidence is capped at 55 because more than half of the seven evidence dimensions are inferred or missing — the conclusion is data-thin.
▸ SRC-CAP-003
Confidence is capped at 65 because none of the seven evidence dimensions reached CONFIRMED quality — there is no high-quality anchor for the analysis.
▸ SRC-CAP-004
Confidence is capped at 65 because telemetry is a single point-in-time snapshot with no ongoing monitoring — right-sizing needs usage patterns over time.
▸ RULE T2 — Right-Sizing Hard Cap
Right-sizing is limited to a moderate 25% reduction because telemetry is a single snapshot. More aggressive right-sizing needs 30+ days of monitoring data.
Decision Path
The step-by-step trace of how the recommendation was reached, in the system's own terms.
PRE-CHECK: hardware_dependency=false, life_safety_classification=UNKNOWN (not true) → pre-check does not fire. RULE 1 (REPURCHASE): saas_available=true ✓, saas_confidence_inferred=72 ≥ 50 ✓, hardware_dependency=false (not true) ✓ → RULE 1 FIRES. strategic_recommendation=REPURCHASE. CONFIDENCE: base=50, telemetry(SINGLE)=+6.0, dependency_mapping(INFERRED)=+2.5, procurement_contracts(INFERRED)=+2.0, architecture_diagrams(MISSING)=+0.0, security_compliance(INFERRED)=+1.25, business_criticality(SINGLE)=+2.0, dr_requirements(MISSING)=+0.0 → pre-cap=63.75. CAP EVALUATION: SRC-CAP-001 does not fire (phi_handled=UNKNOWN, not confirmed true). SRC-CAP-002 fires (5/7 dimensions INFERRED/MISSING → majority) → cap 55. SRC-CAP-003 fires (zero CONFIRMED) → cap 65. SRC-CAP-004 fires (snapshot_count=1, apm_telemetry_available=false) → cap 65. Lowest cap = SRC-CAP-002 at 55. final_confidence=55. GATE: 55 < 60 AND strategic_recommendation=REPURCHASE → recommendation_status=PENDING_CONDITIONS, confidence_gate_applied=true, displayed_primary=REPURCHASE (v4.1: gate does not force RETAIN).
Cap Evaluation (all fired caps): SRC-CAP-002: ≥4 of 7 dimensions INFERRED/MISSING → ceiling 55 (ceiling=55) ← BINDING SRC-CAP-003: Zero CONFIRMED dimensions → ceiling 65 (ceiling=65) SRC-CAP-004: Single snapshot, no APM → ceiling 65 (ceiling=65)
Confidence Scoring Breakdown
How the confidence percentage was calculated, dimension by dimension.
Base score: 50 Evidence contributions: Telemetry SINGLE +6.00 Dependency Mapping INFERRED +2.50 Procurement Contracts INFERRED +2.00 Architecture Diagrams MISSING +0.00 Security Compliance INFERRED +1.25 Business Criticality SINGLE +2.00 Dr Requirements MISSING +0.00 Pre-cap score: 63.75 Binding cap applied: SRC-CAP-002 Final confidence: 55%
Intelligence Gaps
Specific things we don't yet know. Each gap is a reason the confidence score is held back.
BAA confirmation for Siemens Teamplay Cloud
baa_confirmed=UNKNOWN blocks compliant SaaS transition. MUSC must obtain and execute a BAA with Siemens Healthineers for the cloud-hosted platform before any PHI can be transmitted to Teamplay Cloud. Without this, REPURCHASE cannot be executed under HIPAA.
PHI data-flow verification for SaaS platform
phi_handled=UNKNOWN in dependency analysis despite catalog indicating phi_handling=true. A formal PHI data-flow assessment is required to confirm what PHI categories transit through Teamplay, whether Siemens Teamplay Cloud's data residency and processing controls satisfy MUSC's HIPAA/HITRUST obligations, and whether de-identification is feasible.
Contract intelligence for Siemens Teamplay Cloud
contract_intelligence=INSUFFICIENT_EVIDENCE. Current on-premises licensing terms, renewal dates, and SaaS migration pricing are unknown. Without contract data, TCO comparison and migration timeline cannot be confirmed — roi_variance_pct is capped at 60%.
Upstream PACS / imaging infrastructure dependency confirmation
PACS/modality worklist feed is inferred, not confirmed. If Teamplay Cloud requires a different integration pattern (e.g., DICOM push vs. pull, HL7 feed reconfiguration), migration complexity and blast radius may be higher than the current medium estimate.
Architecture diagrams absent
No architecture diagrams available. Integration points, data flows, and network topology are entirely inferred. This prevents accurate migration scoping and cutover planning for the SaaS transition.
DR requirements undefined
No DR/RTO/RPO data available. For a High-criticality clinical analytics platform, DR requirements must be confirmed to validate that Siemens Teamplay Cloud's SLA meets MUSC's operational continuity standards.
APM telemetry unavailable — single RVTools snapshot
apm_telemetry_available=false and snapshot_count=1. Application-layer performance baseline is absent. Right-sizing recommendation (25% reduction) is based on memory active percentage only; actual workload characterization for SaaS sizing cannot be confirmed.
Technical Detail
A fuller technical explanation of the recommendation and its constraints.
Procurement agent confirms saas_available=true with saas_confidence_inferred=72, identifying Siemens Teamplay Cloud as the direct vendor-hosted SaaS replacement — Rule 1 (REPURCHASE) fires. Dependency agent confirms hardware_dependency=false and no latency_sensitive or life_safety constraints that would trigger the structural pre-check or Rule 5. Telemetry agent identifies a single-snapshot moderate utilization profile (31% avg mem active) with a right-sizing opportunity (SRC-RS-002, 25% reduction), consistent with a workload that can be cleanly decommissioned post-SaaS cutover. Confidence is capped at 55 by SRC-CAP-002 (5 of 7 dimensions INFERRED or MISSING), placing the recommendation in PENDING_CONDITIONS status pending BAA confirmation and PHI data-flow verification.
Secondary Path
The fallback option if the primary recommendation cannot proceed.
Secondary Path — RETAIN
RETAIN is the secondary disposition if BAA confirmation cannot be obtained or if PHI data-flow review reveals that Siemens Teamplay Cloud cannot meet MUSC's HIPAA/HITRUST obligations — in that scenario the on-premises deployment would need to persist until a compliant SaaS path is established.
Right-Sizing Notes
Why the cloud sizing estimate is set where it is, based on the telemetry available.
Single point-in-time RVTools snapshot with no APM telemetry. Hard rule T2 (SRC-RS-002) caps maximum tier at MODERATE regardless of observed utilization. fleet_avg_mem_active_pct of 31.0% is independently consistent with MODERATE under T3 thresholds (25%–60%), so no tier conflict exists. CPU utilization data is absent for the sole VM; no estimate substituted per DEF-RS-CPU-NULL. Right-sizing opportunity is flagged at MODERATE (0.25 reduction) but must be validated with multi-snapshot or APM telemetry before any action is taken. Application-type is Client/Server; Rule T1 life-safety override does not apply. OS (Windows Server 2022) is within supported lifecycle; no EOL compliance flags generated.
This tab shows the raw output of each analysis agent — the specialized components that examine telemetry, dependencies, contracts, and cost. It is the most detailed and most technical view. Most readers will find the Overview and Backlog tabs sufficient.
Telemetry Agent
Utilization Profilemoderate
Snapshot Count1
APM AvailableFalse
Hard Rule TriggeredTrue
Right-Sizing Opp.True
Agent Confidence45%
Telemetry — VM Counts
Powered Off VMs0
Flagged VMs0
Mem Active % (avg)31.0%
Data WindowNone
Inventory Source
Dependency Agent
Epic IntegrationFalse
PHI HandledUNKNOWN
Hardware DependencyFalse
Latency SensitiveUNKNOWN
Blast Radiusmedium
Migration RiskMEDIUM
Conf (Inferred)62%
Conf (Confirmed)28%
Verificationsingle_source
Cloud Connectivity
ComplianceHIPAA / HITRUST
Upstream (2)
Imaging / PACS infrastructure (inferred): inferred: Siemens Teamplay is a clinical analytics platform that ingests imaging data; upstream PACS or modality…
Active Directory / Windows authentication (inferred): inferred: application_type=Client/Server on Windows Server 2022; AD integration is standard for enterprise Windows…
Downstream (1)
Clinical analytics consumers / radiology operations (inferred): inferred: Teamplay produces utilization and performance analytics consumed by radiology and operations teams;…
Procurement Agent
SaaS Available—
SaaS VendorSiemens Healthineers
SaaS ProductSiemens Teamplay Cloud
Repurchase ViableINSUFFICIENT_EVIDENCE
BAA ConfirmedUNKNOWN
SaaS Confidence
Inferred Score72%
Confirmed Score0%
Verificationsingle_source
AWS Equiv.None
Provisioning Agent (Python engine — not LLM)
Current VMs1
Recommended VMs1
Right-Sizing25% reduction
Current TCO/yr$6,500
Projected Cloud TCO$5,200
Migration One-Time$1,040
Dual-Running Cost$3,900
-$3,640
Year-1 Net
$1,300
Steady-State
-$1,040
3-Year
AWS TargetAmazon EC2
Azure TargetAzure Virtual Machines
Cost Modeldefault_multipliers
DEF: {'default_id': 'DEF-WL-001', 'field': 'workload_type', 'value': 'app_server', 'reason': 'workload_type indeterminable'}
DEF: {'default_id': 'DEF-MIG-001', 'field': 'migration_factor', 'value': 0.2, 'reason': 'migration_complexity not provided'}
DEF: {'default_id': 'DEF-COST-001', 'field': 'cost_model_basis', 'value': 'default_multipliers', 'reason': 'no customer cloud quote provided; using SRC-TCO-* and SRC-MULT-* registry defaults'}
Portfolio Detail · Discovery Plan
Customer · 36 applications · 587 VMs · Healthcare context
0 of 36
Decommission Opportunity
31 of 36
Migration Candidates
$1,432,900/yr
Run-Rate ROI
(achieved Year 2 onward)
Scenario Bridge
Scenario Bridge — Current State vs. Post-Discovery Target
34 apps are cap-bound — compliance or contract resolution required before migration. 2 apps are structurally blocked (hardware dependency) — RETAIN is the correct long-term posture.
⚙ Computed deterministically — no LLM calls
0
Decommission Opportunity
31
Discovery Sprint Required
5
Structurally RETAIN
Avg Confidence
45% → 57% post-discovery (all 36 apps)
45% → 58% (32 migration-scope apps)
3-Year ROI (5 migration-scope apps)
$1,234,132 (constant; discovery narrows variance, not the point estimate)
Variance: ±55% (current) → ±29% (after discovery)
Discovery Scope
4-6 weeks
across 31 applications (discovery_required + decommission_now)
Philips Perinatal
Structurally RETAIN
Current State
RETAIN
40%
3yr: -$10,400 · ±45% variance
Post-Discovery Target
RETAIN — discovery addresses compliance closure, not migration
34 pts
2wk
Hardware dependency + latency requirement — structurally blo
Philips Patient Monitoring
Structurally RETAIN
Current State
RETAIN
55%
3yr: -$20,800 · ±50% variance
Post-Discovery Target
RETAIN — discovery addresses compliance closure, not migration
29 pts
3wk
Hardware dependency + latency requirement — structurally blo
PaceArt
Structurally RETAIN
Current State
RETAIN
40%
3yr: $13,173 · ±55% variance
Post-Discovery Target
RETAIN — discovery addresses compliance closure, not migration
32 pts
5wk
Confidence ceiling (55%) below gate threshold (60%) — discov
GE Centricity / CCW
Discovery sprint required
Current State
REPURCHASE
40%
3yr: $42,640 · ±60% variance
Post-Discovery Target
REPURCHASE
55%
up from 40%, +15 pts
3yr: $42,640 · ±15% variance
Confirm BAA Status with GE Healthcare foObtain GE Healthcare SaaS Contract Terms
29 pts
3wk
Confidence ceiling (40%) below gate threshold (60%) — discov
Mosaiq
Structurally RETAIN
Current State
REPURCHASE
40%
3yr: $15,253 · ±45% variance
Post-Discovery Target
RETAIN — discovery addresses compliance closure, not migration
15 pts
3wk
Confidence ceiling (40%) below gate threshold (60%) — discov
Siemens Syngo / Syngovia
Discovery sprint required
Current State
REPURCHASE
40%
3yr: $42,640 · ±60% variance
Post-Discovery Target
REPURCHASE
55%
up from 40%, +15 pts
3yr: $42,640 · ±15% variance
Confirm BAA with Siemens Healthineers foValidate Hardware Dependency — DICOM Mod
24 pts
3wk
Confidence ceiling (40%) below gate threshold (60%) — discov
Sectra PACS
Discovery sprint required
Current State
REPURCHASE
40%
3yr: $440,613 · ±60% variance
Post-Discovery Target
REPURCHASE
55%
up from 40%, +15 pts
3yr: $440,613 · ±15% variance
Confirm BAA with Sectra for Medical ClouObtain Sectra Medical Cloud Commercial P
27 pts
3wk
Confidence ceiling (40%) below gate threshold (60%) — discov
Philips IntelliSpace
Discovery sprint required
Current State
REPURCHASE
55%
3yr: $225,333 · ±55% variance
Post-Discovery Target
REPURCHASE
62%
up from 55%, +7 pts
3yr: $225,333 · ±10% variance
Confirm BAA Status with Philips for InteVerify Hardware Dependency for Cardiac I
24 pts
3wk
Confidence ceiling (55%) below gate threshold (60%) — discov
MUSE
Discovery sprint required
Current State
REPURCHASE
40%
3yr: $28,427 · ±50% variance
Post-Discovery Target
REPURCHASE
55%
up from 40%, +15 pts
3yr: $28,427 · ±15% variance
Confirm BAA with GE HealthCare TechnologEngage GE HealthCare Account Team for Sa
36 pts
3wk
Confidence ceiling (40%) below gate threshold (60%) — discov
Lumedx Apollo / CardioPACS
Discovery sprint required
Current State
REPURCHASE
40%
3yr: $15,253 · ±50% variance
Post-Discovery Target
REPURCHASE
55%
up from 40%, +15 pts
3yr: $15,253 · ±15% variance
Obtain signed BAA with Lumedx covering cEngage Lumedx procurement for SaaS prici
21 pts
3wk
Confidence ceiling (40%) below gate threshold (60%) — discov
Hologic Apex
Structurally RETAIN
Current State
RETAIN
55%
3yr: -$2,080 · ±45% variance
Post-Discovery Target
RETAIN — discovery addresses compliance closure, not migration
21 pts
3wk
Confidence ceiling (55%) below gate threshold (60%) — discov
Acuo VNA
Discovery sprint required
Current State
REPURCHASE
40%
3yr: $42,640 · ±50% variance
Post-Discovery Target
REPURCHASE
55%
up from 40%, +15 pts
3yr: $42,640 · ±15% variance
Confirm BAA with Hyland covering Acuo ClObtain current Acuo Technologies contrac
27 pts
3wk
Confidence ceiling (55%) below gate threshold (60%) — discov
Philips Digital Pathology
Discovery sprint required
Current State
REPURCHASE
40%
3yr: $14,213 · ±45% variance
Post-Discovery Target
REPURCHASE
55%
up from 40%, +15 pts
3yr: $14,213 · ±15% variance
Confirm BAA Status for Philips Digital PConfirm Physical Slide Scanner Hardware
28 pts
2wk
Confidence ceiling (40%) below gate threshold (60%) — discov
RapidAI
Discovery sprint required
Current State
REPURCHASE
40%
3yr: $15,253 · ±40% variance
Post-Discovery Target
REPURCHASE
55%
up from 40%, +15 pts
3yr: $15,253 · ±15% variance
Confirm BAA Status with RapidAI Vendor —Obtain RapidAI Cloud Contract Intelligen
24 pts
3wk
Confidence ceiling (55%) below gate threshold (60%) — discov
EchoPACS
Discovery sprint required
Current State
REHOST
55%
3yr: -$1,040 · ±55% variance
Post-Discovery Target
REHOST
62%
up from 55%, +7 pts
3yr: -$1,040 · ±10% variance
Confirm BAA Status with Cloud Provider aConfirm Hardware Dependency — DICOM Acqu
37 pts
5wk
Confidence ceiling (55%) below gate threshold (60%) — discov
Hyland OnBase / VNA
Discovery sprint required
Current State
REPURCHASE
40%
3yr: $200,027 · ±45% variance
Post-Discovery Target
REPURCHASE
55%
up from 40%, +15 pts
3yr: $200,027 · ±15% variance
Confirm BAA with Hyland for SaaS-hosted Deploy 30-day APM telemetry collection a
39 pts
2wk
Confidence ceiling (40%) below gate threshold (60%) — discov
CardioAccess
Discovery sprint required
Current State
REHOST
55%
3yr: -$1,040 · ±60% variance
Post-Discovery Target
REHOST
62%
up from 55%, +7 pts
3yr: -$1,040 · ±10% variance
Confirm BAA status with Medtronic for CaResolve hardware_dependency and latency_
24 pts
2wk
Confidence ceiling (55%) below gate threshold (60%) — discov
DynaCad Radiology
Discovery sprint required
Current State
REHOST
40%
3yr: -$1,040 · ±60% variance
Post-Discovery Target
REHOST
55%
up from 40%, +15 pts
3yr: -$1,040 · ±15% variance
Confirm BAA with Invivo/Philips for DynaAssess Physical Hardware Dependencies fo
22 pts
2wk
Confidence ceiling (40%) below gate threshold (60%) — discov
Medicom
Discovery sprint required
Current State
REHOST
55%
3yr: -$1,040 · ±55% variance
Post-Discovery Target
REHOST
62%
up from 55%, +7 pts
3yr: -$1,040 · ±10% variance
Confirm BAA Status and PHI Scope with MeVerify Hardware Dependency via On-Site A
22 pts
5wk
Confidence ceiling (55%) below gate threshold (60%) — discov
Infinitt Dental PACS
Discovery sprint required
Current State
REPURCHASE
40%
3yr: $42,640 · ±60% variance
Post-Discovery Target
REPURCHASE
55%
up from 40%, +15 pts
3yr: $42,640 · ±15% variance
Confirm BAA with Infinitt for Cloud PACSObtain Infinitt Cloud PACS contract term
32 pts
3wk
Confidence ceiling (40%) below gate threshold (60%) — discov
ATNA Audit Repository
Discovery sprint required
Current State
REHOST
40%
3yr: $30,507 · ±60% variance
Post-Discovery Target
REHOST
55%
up from 40%, +15 pts
3yr: $30,507 · ±15% variance
Confirm BAA with Sectra for PHI-HandlingValidate Latency Sensitivity of ATNA ARR
34 pts
3wk
Confidence ceiling (40%) below gate threshold (60%) — discov
Butterfly Link
Discovery sprint required
Current State
REPURCHASE
40%
3yr: $15,253 · ±50% variance
Post-Discovery Target
REPURCHASE
55%
up from 40%, +15 pts
3yr: $15,253 · ±15% variance
Execute BAA with Butterfly Network to reObtain Butterfly Cloud contract terms, S
27 pts
3wk
Confidence ceiling (55%) below gate threshold (60%) — discov
HeartFlow Connect
Discovery sprint required
Current State
REPURCHASE
40%
3yr: $14,213 · ±50% variance
Post-Discovery Target
REPURCHASE
55%
up from 40%, +15 pts
3yr: $14,213 · ±15% variance
Confirm Executed BAA with HeartFlow — ReConfirm Life Safety Classification with
23 pts
3wk
Confidence ceiling (55%) below gate threshold (60%) — discov
Vizai
Discovery sprint required
Current State
REPURCHASE
55%
3yr: -$2,080 · ±55% variance
Post-Discovery Target
REPURCHASE
62%
up from 55%, +7 pts
3yr: -$2,080 · ±10% variance
Confirm BAA with Viz.ai for Viz.ai CloudObtain Viz.ai Cloud Procurement Contract
43 pts
3wk
Confidence ceiling (55%) below gate threshold (60%) — discov
GE Unified Viewer
Discovery sprint required
Current State
REHOST
55%
3yr: $27,387 · ±55% variance
Post-Discovery Target
REHOST
62%
up from 55%, +7 pts
3yr: $27,387 · ±10% variance
Confirm BAA Status with GE HealthCare TeEngage GE HealthCare to Confirm Cloud Ho
18 pts
5wk
Confidence ceiling (55%) below gate threshold (60%) — discov
LifeImage / Expedo
Discovery sprint required
Current State
REPURCHASE
55%
3yr: $14,213 · ±60% variance
Post-Discovery Target
REPURCHASE
64%
up from 55%, +9 pts
3yr: $14,213 · ±10% variance
Confirm BAA with Hyland for NilRead ClouValidate DICOM Routing and Image Exchang
42 pts
3wk
Confidence ceiling (55%) below gate threshold (60%) — discov
Magview
Discovery sprint required
Current State
REPURCHASE
40%
3yr: $13,173 · ±50% variance
Post-Discovery Target
REPURCHASE
65%
up from 40%, +25 pts
3yr: $13,173 · ±15% variance
Confirm BAA with Magview Cloud SaaS VendObtain Magview Cloud Contract Terms, ACV
24 pts
2wk
Confidence ceiling (40%) below gate threshold (60%) — discov
PenRad
Discovery sprint required
Current State
REPURCHASE
40%
3yr: -$2,080 · ±40% variance
Post-Discovery Target
REPURCHASE
65%
up from 40%, +25 pts
3yr: -$2,080 · ±15% variance
Confirm BAA with PenRad Technologies forObtain PenRad Cloud SaaS contract terms,
32 pts
5wk
Confidence ceiling (40%) below gate threshold (60%) — discov
Circle
Discovery sprint required
Current State
REPURCHASE
40%
3yr: -$1,040 · ±60% variance
Post-Discovery Target
REPURCHASE
55%
up from 40%, +15 pts
3yr: -$1,040 · ±15% variance
Confirm BAA Coverage for Circle CVi ClouValidate PACS/RIS Integration Compatibil
34 pts
3wk
Confidence ceiling (40%) below gate threshold (60%) — discov
Nuance PowerShare
Discovery sprint required
Current State
REPURCHASE
40%
3yr: $26,347 · ±60% variance
Post-Discovery Target
REPURCHASE
55%
up from 40%, +15 pts
3yr: $26,347 · ±15% variance
Confirm BAA Status for Nuance PowerShareObtain Current Licensing Contract and Sa
32 pts
3wk
Confidence ceiling (55%) below gate threshold (60%) — discov
Nuance PowerScribe One
Discovery sprint required
Current State
REPURCHASE
40%
3yr: -$1,040 · ±50% variance
Post-Discovery Target
REPURCHASE
55%
up from 40%, +15 pts
3yr: -$1,040 · ±15% variance
Confirm BAA Coverage — Nuance PowerScribRetrieve Current On-Premises License Con
26 pts
5wk
Confidence ceiling (40%) below gate threshold (60%) — discov
Optellum / Lung AI
Discovery sprint required
Current State
REPURCHASE
55%
3yr: -$2,080 · ±60% variance
Post-Discovery Target
REPURCHASE
62%
up from 55%, +7 pts
3yr: -$2,080 · ±10% variance
Execute HIPAA BAA with Optellum for OpteObtain and Review Existing Optellum Cont
25 pts
3wk
Confidence ceiling (55%) below gate threshold (60%) — discov
cNeuro
Discovery sprint required
Current State
REHOST
55%
3yr: -$1,040 · ±60% variance
Post-Discovery Target
REHOST
64%
up from 55%, +9 pts
3yr: -$1,040 · ±10% variance
Confirm BAA Status with Cortechs.ai and Classify PHI Handling Scope and Confirm
27 pts
2wk
Confidence ceiling (55%) below gate threshold (60%) — discov
Volpara
Discovery sprint required
Current State
REPURCHASE
40%
3yr: -$1,040 · ±60% variance
Post-Discovery Target
REPURCHASE
55%
up from 40%, +15 pts
3yr: -$1,040 · ±15% variance
Confirm BAA with Volpara Health for VolpObtain Volpara Health Cloud SaaS contrac
20 pts
3wk
Confidence ceiling (40%) below gate threshold (60%) — discov
ACRConnect
Discovery sprint required
Current State
REPURCHASE
55%
3yr: -$1,040 · ±55% variance
Post-Discovery Target
REPURCHASE
62%
up from 55%, +7 pts
3yr: -$1,040 · ±10% variance
Confirm PHI Handling Classification via Obtain Executed BAA with ACR Covering AC
17 pts
3wk
Confidence ceiling (55%) below gate threshold (60%) — discov
Teamplay
Discovery sprint required
Current State
REPURCHASE
55%
3yr: -$1,040 · ±60% variance
Post-Discovery Target
REPURCHASE
64%
up from 55%, +9 pts
3yr: -$1,040 · ±10% variance
Confirm BAA status with Siemens HealthinConduct formal PHI data-flow assessment
22 pts
3wk
Confidence ceiling (55%) below gate threshold (60%) — discov
* Timelines reflect end-to-end calendar weeks. Items run in parallel; total elapsed time is bounded by the longest single workstream.
Discovery Sprint Plan
AppWhat to doOwnerTimelineUnlocks
ACRConnect Verify whether ACRConnect stores or transmits patient health information Compliance Officer / Application Owner (Erika Murphy) 7 days compliance_close, cap_resolution
Philips Digital Pathology Verify Business Associate Agreement with Philips for Cloud Migration Compliance Officer / Legal 10 days compliance_close, cap_resolution, migration_commit
Philips Digital Pathology Determine Whether Physical Scanners Are Tied to the On-Premises Servers Infrastructure Lead / Philips Field Engineer 10 days migration_commit, operational_safety
CardioAccess Determine whether CardioAccess is classified as a life-safety system by clinical leadership CMIO / Clinical Informatics 5 days operational_safety, migration_commit, cap_resolution
DynaCad Radiology Verify Business Associate Agreement with DynaCad Vendor Before Cloud Migration Healthcare Compliance Officer / Legal Counsel 14 days compliance_close, cap_resolution, migration_commit
DynaCad Radiology Determine Whether DynaCad Requires Physical Hardware to Operate Infrastructure Architect / Vendor Engagement Lead 10 days migration_commit, operational_safety
Medicom Verify Business Associate Agreement and Patient Data Handling Before Any Cloud Move Compliance Officer / Legal Counsel 21 days compliance_close, cap_resolution, migration_commit
Medicom Confirm Whether Medicom Connects to Physical Radiology Hardware That Would Block Cloud Migration Application Owner (Wendy Ketchum, Radiology Informatics) + Infrastructure Engineer 14 days migration_commit, cap_resolution, operational_safety
HeartFlow Connect Verify Whether HeartFlow Connect Is Used in Real-Time Clinical Decision Support Cardiology Application Owner / Clinical Informatics 10 days migration_commit, operational_safety
Magview Obtain Business Associate Agreement from Magview for Cloud Product Compliance Officer / Legal Counsel 10 days compliance_close, cap_resolution, migration_commit
Magview Collect SaaS Pricing and Contract Details from Magview Vendor Procurement / Vendor Management 14 days cost_clarity, migration_commit
PenRad Verify Business Associate Agreement with PenRad Technologies MUSC Legal / Compliance Officer 21 days compliance_close, cap_resolution, migration_commit
Circle Install Performance Monitoring on the Circle CVI Server for 30 Days Infrastructure / Platform Engineering 35 days cap_resolution, cost_clarity
Nuance PowerShare Verify Business Associate Agreement for Cloud Migration Legal / Compliance Officer + Microsoft Account Manager 10 days cap_resolution, compliance_close, migration_commit
Nuance PowerShare Get Current Contract Terms and Cloud Pricing for PowerShare Procurement / Microsoft Account Manager 15 days cost_clarity, migration_commit
Optellum / Lung AI Confirm HIPAA Business Associate Agreement with Optellum before any cloud migration Compliance Officer / Legal 10 days compliance_close, cap_resolution, migration_commit
cNeuro Verify Business Associate Agreement for cNeuro Compliance Officer / Legal Counsel 14 days compliance_close, cap_resolution, migration_commit
cNeuro Confirm Whether cNeuro Handles Patient Data and Whether It Is Live in Production Application Owner / HIPAA Privacy Officer 10 days compliance_close, cap_resolution, migration_commit
ACRConnect Confirm a signed data privacy agreement with ACR for the cloud product Legal / Compliance Officer 21 days compliance_close, cap_resolution, migration_commit
ACRConnect Get pricing and contract details for ACRConnect Cloud from the vendor Procurement / Vendor Manager 21 days cost_clarity, migration_commit, cap_resolution
Lumedx Apollo / CardioPACS Confirm Business Associate Agreement with Lumedx for cloud migration Legal / Compliance Officer 10 days compliance_close, cap_resolution
Lumedx Apollo / CardioPACS Confirm whether CardioPACS has physical hardware connections that would block cloud migration Clinical Informatics / Infrastructure Architect 10 days migration_commit, operational_safety
ATNA Audit Repository Confirm Whether Real-Time Audit Logging Has Strict Timing Requirements Radiology Application Owner (Wendy Ketchum) / Network Architect 10 days migration_commit, operational_safety
GE Unified Viewer Verify Business Associate Agreement Coverage for Cloud Migration Compliance Officer / Legal 10 days compliance_close, cap_resolution, migration_commit
GE Unified Viewer Confirm with GE HealthCare Whether Unified Viewer Can Be Hosted in the Cloud Vendor Management / IT Procurement 14 days migration_commit, cap_resolution, cost_clarity
Volpara Verify Business Associate Agreement for Volpara Cloud Migration Compliance Officer / Legal 10 days compliance_close, cap_resolution, migration_commit
Teamplay Verify Business Associate Agreement for Siemens Teamplay Cloud Compliance Officer / Legal 10 days compliance_close, cap_resolution, migration_commit
LifeImage / Expedo Verify Business Associate Agreement with Hyland for Cloud Migration Compliance Officer / Procurement Lead 14 days compliance_close, cap_resolution, migration_commit
Mosaiq Obtain Signed Business Associate Agreement with Elekta for Cloud Product Compliance Officer / Legal Counsel 10 days cap_resolution, compliance_close, migration_commit
Mosaiq Determine Whether Real-Time Speed Is Required for Radiation Treatment Delivery RAD/ONC Physics / Clinical Engineering (Jim Early / Nayquan Scott) 10 days migration_commit, operational_safety
Siemens Syngo / Syngovia Confirm Business Associate Agreement with Siemens Healthineers for Cloud Migration Compliance Officer / Legal 10 days compliance_close, cap_resolution
Acuo VNA Retrieve Current Contract Terms and Hyland Cloud Pricing Proposal Procurement Manager / Vendor Relationship Manager 14 days cost_clarity, migration_commit
Vizai Confirm Whether Patient Health Information Is Processed by Viz.ai Data Owner / HIPAA Privacy Officer 7 days compliance_close, cap_resolution
EchoPACS Confirm Data Privacy Agreements Before Any Cloud Migration of Patient Imaging Data Legal / Compliance Officer 21 days compliance_close, cap_resolution, migration_commit
EchoPACS Determine Whether Echo Ultrasound Machines Require On-Site Server Proximity Infrastructure Architect / Adult Cardiology IT 14 days migration_commit, cap_resolution, operational_safety
EchoPACS Collect 30 Days of Performance Data to Validate Cloud Sizing and Lift Telemetry Cap Infrastructure / Monitoring Team 30 days cap_resolution, cost_clarity, migration_commit
CardioAccess Verify Business Associate Agreement with Medtronic before any cloud migration Legal / Compliance Officer 10 days compliance_close, cap_resolution, migration_commit
CardioAccess Determine whether CardioAccess connects to physical cardiac devices or requires low-latency network links Enterprise Architect / Clinical Engineering 10 days migration_commit, cap_resolution, operational_safety
Medicom Obtain Medicom Contract Details and Ask Vendor About Cloud or SaaS Options Procurement / Vendor Management 30 days cost_clarity, cap_resolution, migration_commit
Butterfly Link Obtain Business Associate Agreement with Butterfly Network Compliance / Legal 21 days compliance_close, cap_resolution, migration_commit
HeartFlow Connect Obtain Signed Business Associate Agreement with HeartFlow to Unblock Migration Decision Compliance Officer / Legal Counsel 21 days compliance_close, cap_resolution, migration_commit
Magview Confirm That Magview Cloud Supports the Existing MyChart Integration Integration Architect / Vendor Engagement Lead 21 days migration_commit, operational_safety
PenRad Get PenRad Cloud Pricing, SLA, and Migration Terms from Vendor Procurement / Vendor Management 30 days cost_clarity, migration_commit
Circle Obtain Business Associate Agreement for Circle CVi Cloud SaaS Compliance Officer / Legal Counsel 21 days cap_resolution, compliance_close, migration_commit
Circle Confirm Radiology System Integration Works with Circle CVi Cloud Enterprise Architect / Radiology IT Lead 21 days migration_commit, operational_safety
Nuance PowerScribe One Confirm Business Associate Agreement for Nuance Cloud Product Compliance Officer / Legal Counsel 30 days compliance_close, cap_resolution, migration_commit
Nuance PowerScribe One Pull Existing Nuance License Contract and Renewal Terms Procurement / Contract Manager 21 days cost_clarity, migration_commit
Optellum / Lung AI Pull the current Optellum contract and get SaaS pricing before committing to migration Procurement / Vendor Management 15 days cost_clarity, migration_commit, cap_resolution
cNeuro Install Application Performance Monitoring on cNeuro Server Infrastructure Engineer / Cloud Migration Team 35 days cap_resolution, cost_clarity, migration_commit
Lumedx Apollo / CardioPACS Get Lumedx cloud pricing and contract details to validate the SaaS migration path Procurement / Vendor Management 21 days migration_commit, cost_clarity, cap_resolution
RapidAI Confirm Business Associate Agreement with RapidAI Privacy/Compliance Officer + Vendor Management 15 days compliance_close, cap_resolution, migration_commit
RapidAI Gather RapidAI Cloud Contract and Pricing Terms Procurement / Vendor Management 15 days cost_clarity, migration_commit
Infinitt Dental PACS Obtain Business Associate Agreement with Infinitt for Cloud PACS Compliance Officer / Legal Counsel 21 days cap_resolution, compliance_close, migration_commit
Infinitt Dental PACS Request and Review Infinitt Cloud PACS Contract and Pricing Proposal Procurement / Vendor Management 21 days cost_clarity, migration_commit
Infinitt Dental PACS Map All System Connections for Infinitt Dental PACS Enterprise Architect / Integration Team 14 days migration_commit, operational_safety
ATNA Audit Repository Verify Business Associate Agreement with Sectra Covers Cloud Hosting Compliance Officer / Legal Counsel 14 days compliance_close, cap_resolution, migration_commit
ATNA Audit Repository Install Application Performance Monitoring to Validate Server Sizing Infrastructure / Platform Engineering 21 days cost_clarity, migration_commit
GE Unified Viewer Collect 30 Days of Performance Data to Validate Right-Sizing Recommendation Infrastructure / Cloud Engineering 30 days cap_resolution, cost_clarity, operational_safety
Volpara Secure SaaS Pricing and Contract Terms from Volpara Health Procurement / Vendor Management 21 days cost_clarity, migration_commit
Teamplay Map and verify PHI data flows for Teamplay on-premises and cloud platforms Security / HIPAA Privacy Officer 15 days compliance_close, cap_resolution, migration_commit
Teamplay Document Teamplay system architecture and integration points Enterprise Architect / Application Owner 10 days cap_resolution, migration_commit, operational_safety
Hyland OnBase / VNA Confirm Business Associate Agreement with Hyland for Cloud Hosting Compliance Officer / Legal Counsel 10 days compliance_close, cap_resolution, migration_commit
Hyland OnBase / VNA Get Hyland Cloud Pricing and Contract Terms for SaaS Migration Procurement / Vendor Management 10 days cost_clarity, migration_commit
LifeImage / Expedo Document DICOM Image Routing and Integration Architecture Enterprise Architect / Radiology Informatics Lead 21 days migration_commit, cap_resolution, operational_safety
GE Centricity / CCW Verify the Business Associate Agreement with GE Healthcare before any migration can proceed Compliance Officer / Legal Counsel 21 days compliance_close, cap_resolution, migration_commit
GE Centricity / CCW Get formal pricing and contract terms from GE Healthcare for the cloud version of Centricity Procurement / Vendor Management 21 days cost_clarity, migration_commit
Siemens Syngo / Syngovia Confirm Whether Syngo Connects to Physical Imaging Hardware (CT, MRI, Cath Lab) Clinical Informatics Lead / Infrastructure Architect 20 days migration_commit, operational_safety
Acuo VNA Verify Business Associate Agreement with Hyland for Cloud VNA Migration Compliance Officer / Legal Counsel 21 days compliance_close, cap_resolution, migration_commit
Mosaiq Confirm Whether Mosaiq Is Physically Wired to the Linear Accelerator RAD/ONC Clinical Engineering Lead (Jim Early / Nayquan Scott) 21 days migration_commit, operational_safety
Vizai Verify Business Associate Agreement with Viz.ai for Cloud Migration Legal / Compliance Officer 21 days compliance_close, cap_resolution, migration_commit
Vizai Secure Formal Contract and Pricing Proposal for Viz.ai Cloud Procurement / Vendor Management 21 days cost_clarity, cap_resolution, migration_commit
Philips IntelliSpace Verify Business Associate Agreement with Philips for Cloud Platform Legal / Compliance Officer 10 days compliance_close, cap_resolution, migration_commit
Philips IntelliSpace Confirm That IntelliSpace Processes Patient Health Information Privacy Officer / Application Owner 10 days compliance_close, cap_resolution
MUSE Verify Business Associate Agreement with GE HealthCare — Required Before Any Migration Can Proceed Compliance Officer / Legal 21 days compliance_close, cap_resolution
MUSE Get Formal Pricing and Migration Proposal from GE HealthCare for Cloud-Hosted MUSE Procurement / Vendor Management 21 days cost_clarity, migration_commit
Sectra PACS Verify Business Associate Agreement with Sectra for Cloud Hosting Compliance Officer / Legal Counsel 10 days compliance_close, cap_resolution
Hyland OnBase / VNA Collect 30-Day Performance Monitoring Data Across All 55 VMs Infrastructure / Cloud Engineering Team 30 days cost_clarity, cap_resolution
Philips IntelliSpace Confirm Whether Cardiac Imaging Hardware Requires On-Premises Connectivity Enterprise Architect / Cardiology IT Lead 20 days migration_commit, cap_resolution, operational_safety
Sectra PACS Get Sectra Cloud Pricing and Contract Details Procurement / Vendor Management 21 days cost_clarity, migration_commit
Compliance Closure Plan (Structurally RETAIN apps)
These items address regulatory close-out, not migration prep.
AppWhat to doOwnerTimelineUnlocks
Hologic Apex Get a Firm Date and Target Version for the In-Progress Windows Server Upgrade Mike Slezak (Application Owner) + Cloud Migration Lead 14 days migration_commit, cap_resolution, operational_safety
Hologic Apex Confirm a Business Associate Agreement Is in Place Before Any Cloud Migration Legal / Compliance Officer + Cloud Procurement Lead 14 days compliance_close, migration_commit
Philips Perinatal Confirm Business Associate Agreement with Philips for Perinatal System Compliance Officer / Procurement Lead 10 days cap_resolution, compliance_close
Mosaiq Obtain Signed Business Associate Agreement with Elekta for Cloud Product Compliance Officer / Legal Counsel 10 days cap_resolution, compliance_close, migration_commit
Mosaiq Determine Whether Real-Time Speed Is Required for Radiation Treatment Delivery RAD/ONC Physics / Clinical Engineering (Jim Early / Nayquan Scott) 10 days migration_commit, operational_safety
Philips Patient Monitoring Locate or execute a Business Associate Agreement with Philips for patient monitoring data Compliance Officer / Legal Counsel 10 days compliance_close, cap_resolution
Philips Patient Monitoring Pull the current Philips support contract to understand renewal dates and costs Procurement / Vendor Management 10 days cost_clarity, migration_commit
Hologic Apex Determine Whether Hologic Apex Is Physically Tethered to On-Site DEXA Scanners Infrastructure Architect + Mike Slezak (Application Owner) 21 days migration_commit, cap_resolution
Philips Perinatal Install Performance Monitoring on Perinatal Servers for 30 Days Infrastructure Engineer / Biomedical IT 10 days cap_resolution, cost_clarity
Philips Perinatal Obtain Current Philips Perinatal Contract Details Procurement Lead 10 days cost_clarity, migration_commit
Mosaiq Confirm Whether Mosaiq Is Physically Wired to the Linear Accelerator RAD/ONC Clinical Engineering Lead (Jim Early / Nayquan Scott) 21 days migration_commit, operational_safety
PaceArt Confirm Business Associate Agreement for PaceArt PHI Handling Compliance Officer / Legal Counsel 21 days compliance_close, cap_resolution
PaceArt Get Biotronik's Official Position on Hardware Requirements and Cloud Compatibility Application Owner / Vendor Management 30 days migration_commit, operational_safety
Philips Patient Monitoring Verify which physical bedside devices are directly connected to the Philips monitoring servers Clinical Engineering / Philips Vendor POC (Matthew Corson) 15 days migration_commit, operational_safety
Defaults Used
This analysis used 69 default assumptions. The defaults are documented and replaceable with client-provided data.
View Defaults Glossary (top 5 most common)
DEF-COST-001 — Cloud cost estimate: When no vendor quote is available, cloud TCO is estimated using AWS on-demand list pricing for the recommended instance family, adjusted by the right-sizing reduction factor. This is a directional estimate; actual costs depend on reserved instance commitments, EDP discounts, and data transfer patterns.
DEF-MIG-001 — Migration one-time cost: Migration labor is estimated at 40 hours per VM at a blended rate of $150/hr when no SOW or vendor quote is provided. This covers lift-and-shift effort only; refactor or replatform work requires a separate estimate.
DEF-WL-001 — Workload sizing: When APM telemetry is unavailable, right-sizing is capped at a 25% vCPU/RAM reduction (SRC-RS-002 MODERATE confidence). Continuous telemetry over 30+ days would enable more aggressive right-sizing.
DEF-BAA-001 — BAA status: When no BAA artifact is on file, baa_confirmed is set to UNKNOWN. This triggers SRC-CAP-001 (confidence cap 40) for any PHI-handling application. The cap is lifted only when a compliance officer provides written attestation.
DEF-DR-001 — DR/RTO requirements: When no DR documentation is available, RTO/RPO requirements are assumed to match the application criticality tier: High = 4hr RTO / 1hr RPO, Medium = 24hr / 4hr, Low = 72hr / 24hr. Actual requirements must be confirmed before migration wave planning.
Validator Status
✓ All agents PASS or PASS_WITH_DEFAULTS
Apptelemetrydependencyprocurementprovisioningsynthesizerconfidence_advisor
Philips PerinatalPASSPASSPASSPASS_WITH_DEFAULTSPASSPASS
Philips Patient MonitoringPASSPASSPASSPASS_WITH_DEFAULTSPASSPASS
PaceArtPASSPASSPASSPASS_WITH_DEFAULTSPASSPASS
GE Centricity / CCWPASSPASSPASSPASS_WITH_DEFAULTSPASSPASS
MosaiqPASSPASSPASSPASS_WITH_DEFAULTSPASSPASS
Siemens Syngo / SyngoviaPASSPASSPASSPASS_WITH_DEFAULTSPASSPASS
Sectra PACSPASSPASSPASSPASS_WITH_DEFAULTSPASSPASS
Philips IntelliSpacePASSPASSPASSPASS_WITH_DEFAULTSPASSPASS
MUSEPASSPASSPASSPASS_WITH_DEFAULTSPASSPASS
Lumedx Apollo / CardioPACSPASSPASSPASSPASS_WITH_DEFAULTSPASSPASS
Hologic ApexPASSPASSPASSPASS_WITH_DEFAULTSPASSPASS
Acuo VNAPASSPASSPASSPASS_WITH_DEFAULTSPASSPASS
Philips Digital PathologyPASSPASSPASSPASS_WITH_DEFAULTSPASSPASS
RapidAIPASSPASSPASSPASS_WITH_DEFAULTSPASSPASS
EchoPACSPASSPASSPASSPASS_WITH_DEFAULTSPASSPASS
Hyland OnBase / VNAPASSPASSPASSPASS_WITH_DEFAULTSPASSPASS
CardioAccessPASSPASSPASSPASS_WITH_DEFAULTSPASSPASS
DynaCad RadiologyPASSPASSPASSPASS_WITH_DEFAULTSPASSPASS
MedicomPASSPASSPASSPASS_WITH_DEFAULTSPASSPASS
Infinitt Dental PACSPASSPASSPASSPASS_WITH_DEFAULTSPASSPASS
ATNA Audit RepositoryPASSPASSPASSPASS_WITH_DEFAULTSPASSPASS
Butterfly LinkPASSPASSPASSPASS_WITH_DEFAULTSPASSPASS
HeartFlow ConnectPASSPASSPASSPASS_WITH_DEFAULTSPASSPASS
VizaiPASSPASSPASSPASS_WITH_DEFAULTSPASSPASS
GE Unified ViewerPASSPASSPASSPASS_WITH_DEFAULTSPASSPASS
LifeImage / ExpedoPASSPASSPASSPASS_WITH_DEFAULTSPASSPASS
MagviewPASSPASSPASSPASS_WITH_DEFAULTSPASSPASS
PenRadPASSPASSPASSPASS_WITH_DEFAULTSPASSPASS
CirclePASSPASSPASSPASS_WITH_DEFAULTSPASSPASS
Nuance PowerSharePASSPASSPASSPASS_WITH_DEFAULTSPASSPASS
Nuance PowerScribe OnePASSPASSPASSPASS_WITH_DEFAULTSPASSPASS
Optellum / Lung AIPASSPASSPASSPASS_WITH_DEFAULTSPASSPASS
cNeuroPASSPASSPASSPASS_WITH_DEFAULTSPASSPASS
VolparaPASSPASSPASSPASS_WITH_DEFAULTSPASSPASS
ACRConnectPASSPASSPASSPASS_WITH_DEFAULTSPASSPASS
TeamplayPASSPASSPASSPASS_WITH_DEFAULTSPASSPASS
▶ Pipeline Stats — for analyst review
36
Applications
408
Total VMs
45.0%
Avg Confidence
32
Pending Conditions
36
Caps Applied
0
Validator Failures
69
Defaults Used
Methodology · Basis · Assumptions

R-ARF — How This Analysis Works

R-ARF (Rationalization Automation and Recommendation Framework) analyzes your application portfolio and produces 6R cloud migration recommendations. Every number on this report traces to evidence, every confidence score is bounded by documented caps, and every dollar value is computed by a Python engine — never by an AI model.

This page documents the framework so you can verify any recommendation against the methodology that produced it. Six short sections cover the recommendations, confidence, savings, risk, governance, and worked examples for each calculation.

Estimated read time: 6 minutes.

The Recommendation

R-ARF analyzes each application and recommends one of five paths. We pick the path that makes the most strategic sense given the evidence, with RETAIN reserved for apps that physically cannot move.

REPURCHASE
Replace with vendor SaaS. The cleanest path when the vendor offers an equivalent cloud service.
RETIRE
Decommission, often into a successor system already in the portfolio. The fastest immediate cost recovery.
REFACTOR
Rewrite as cloud-native (serverless, managed services). Reserved for internally developed apps where the modernization investment pays back.
REHOST
Lift-and-shift to cloud VMs. Lowest-risk migration when SaaS isn't available and refactor isn't warranted.
RETAIN
Keep on-premises. Reserved for apps with hardware integration, latency requirements, or life-safety classifications that physically cannot move to cloud.

We evaluate the five rules in priority order. The first rule that matches wins. RETAIN is the last resort, not the default — it only fires when an app structurally cannot migrate.

For each app's specific reasoning, see the Decision Path on its App Detail page.

How We Score Confidence

Every recommendation comes with a confidence score from 0 to 100. The score reflects how much evidence we have, not how strongly we endorse the recommendation. Above 60, the recommendation is ready to execute. Below 60, we've identified a viable path but specific data gaps must close before committing.

The score is built from seven evidence dimensions:

  • Telemetry — VM utilization data, snapshot counts, APM coverage
  • Dependency Mapping — Upstream/downstream system relationships
  • Procurement Contracts — Vendor agreements, BAAs, renewal terms
  • Architecture Diagrams — Interface and integration documentation
  • Security Compliance — HIPAA, HITRUST, SOC2 posture
  • Business Criticality — Clinical impact, revenue dependency
  • DR Requirements — Recovery objectives, business continuity plans
Evidence qualityMultiplierWhat it means
CONFIRMED1.0Multiple sources agree
SINGLE0.5One source, no contradiction
INFERRED0.25Inferred from related data
MISSING0.0No evidence available

The score starts at a base of 50 and can climb up to 50 more depending on evidence quality across the seven dimensions.

PHILIPS PERINATAL — CONFIDENCE SCORE WORKED EXAMPLE

Start with the base score:                                    50

Add evidence contributions across 7 dimensions:
  Dimension                    Evidence     Multiplier   Contribution
  ─────────────────────────────────────────────────────────────────────
  Telemetry                    SINGLE       0.5          +6
  Dependency Mapping           SINGLE       0.5          +5
  Procurement Contracts        MISSING      0.0          +0
  Architecture Diagrams        INFERRED     0.25         +1.75
  Security Compliance          INFERRED     0.25         +1.25
  Business Criticality         CONFIRMED    1.0          +4
  DR Requirements              INFERRED     0.25         +1
                                                              ─────
                                                 Subtotal:   +19.0

Pre-cap score:                                               69.0

Apply confidence caps:
  Unconfirmed BAA cap (PHI + UNKNOWN BAA)            ceiling: 40
  Heavy inferred evidence (>50% inferred)            ceiling: 55

Lowest cap wins (most restrictive):                          40

Final confidence:                                            40%

Reading the example: the evidence is strong overall — pre-cap score of 73 reflects solid dependency mapping and a clear migration path. But the unconfirmed BAA caps the displayed confidence at 40 until that gap closes. Confirming the BAA lifts the cap to 55 (the next-binding cap). Subsequent discovery work lifts further.

For each app's seven-dimension breakdown, see the Provenance tab on its App Detail page.

What "Data Quality" Means

Alongside the confidence score, each application shows a Data Quality percentage. The two measure different things:

  • Confidence — how sure we are of the recommendation, after caps. This is the primary decision signal.
  • Data Quality — how complete and well-sourced the underlying input data is, independent of the recommendation. It reflects how many of the evidence dimensions are backed by CONFIRMED or SINGLE-source evidence versus INFERRED or MISSING.

An app can have low confidence but moderate data quality (the evidence is reasonable but a cap binds), or the reverse. Data Quality tells you how much the analysis rested on assumptions; Confidence tells you how firm the recommendation is. The Data Quality score is computed as: (sum of dimension weights for CONFIRMED + SINGLE dimensions) ÷ 50 × 100.

Why Some Apps Stay Below 60

Some apps have specific gaps that cap confidence regardless of how strong other evidence is. These caps are intentional — they protect the recommendation from over-confidence in the presence of unresolved risks.

  • Unconfirmed Business Associate Agreement (BAA) — For an app handling PHI, confidence caps at 40 because PHI transfer to a cloud vendor cannot be authorized without confirmed BAA coverage.
  • Limited telemetry — A single VM snapshot without ongoing APM monitoring caps confidence at 65 because right-sizing decisions rely on usage patterns, not point-in-time samples.
  • Heavy reliance on inferred evidence — When more than half of the seven dimensions are INFERRED or MISSING, confidence caps at 55 because the conclusion is data-thin even if the data is consistent.
  • No confirmed evidence at all — When none of the seven dimensions reach CONFIRMED quality, confidence caps at 65 because there is no high-quality anchor for the analysis.

When multiple caps apply, the lowest cap binds. Resolving the binding cap (typically by completing one or more action items in the app's Backlog) lifts confidence to the next-binding cap, and so on.

PHILIPS PERINATAL — CONFIDENCE PROGRESSION AS CAPS RESOLVE

Step 0 — Current state
  Pre-cap score:                                69.0
  Binding cap: unconfirmed BAA                  ceiling: 40
  Displayed confidence:                         40%
  Status: PENDING CONDITIONS (below 60 threshold)

Step 1 — BAA novation confirmed
  Pre-cap score:                                69.0 (unchanged)
  Previous binding cap lifts (BAA confirmed)
  New binding cap: heavy inferred evidence      ceiling: 55
  Displayed confidence:                         55%
  Status: still PENDING CONDITIONS (below 60)

Step 2 — APM telemetry deployed (30 days of data)
  Telemetry dimension upgrades: SINGLE → CONFIRMED
  Pre-cap score updated:                        75.0
  Caps that still apply: heavy inferred         ceiling: 55
  Displayed confidence:                         55% (cap still binds)

Step 3 — Contract terms retrieved
  Procurement Contracts: INFERRED → CONFIRMED
  Pre-cap score updated:                        79.0
  3 of 7 dims INFERRED/MISSING — below the majority (4 of 7) threshold
  No remaining caps below pre-cap
  Displayed confidence:                         79.0%
  Status: EXECUTABLE (above 60 threshold)

Reading the example: each backlog item lifts a specific cap or upgrades a specific evidence dimension. The displayed confidence climbs as caps resolve and evidence quality improves. The recommendation becomes EXECUTABLE when confidence crosses 60.

For technical cap identifiers (SRC-CAP-001 through SRC-CAP-004) and their precise trigger conditions, see the Decision Path on each App Detail page.

How We Compute Savings

Annual savings equal current on-premises cost minus projected cloud cost. We use $6,500 per VM per year as the on-premises baseline — an industry-standard figure that covers hardware, operating system licenses, hosting, and operational overhead. Cloud cost is typically 60-80% of on-premises due to right-sizing and elasticity benefits.

Year 1 net is usually negative because migration has up-front costs: one-time migration work plus dual-running both environments during cutover. Year 2 onward captures the full annual savings — what we call the "Run-Rate ROI" headline number.

PHILIPS PERINATAL — WORKED EXAMPLE

Current on-prem cost:    10 VMs × $6,500/yr  =  $65,000/yr
Projected cloud cost:    10 VMs × $5,200/yr  =   $52,000/yr
                                                ─────────
Steady-state savings:    $65,000 - $52,000  =   $13,000/yr

Year 1 net:              $13,000 savings
                       - $10,400 migration one-time
                       - $39,000 dual-running
                                                ─────────
                                                -$36,400

Year 2 onward:           +$13,000/yr (full run-rate)

3-Year cumulative:       -$36,400 + $13,000 + $13,000  =  +$-10,400
MUSC PORTFOLIO — RUN-RATE ROI ROLLUP

Migrating apps (REPURCHASE / REFACTOR):
  GE Centricity / CCW            REPURCHASE     +$31,200/yr
  Siemens Syngo / Syngovia       REPURCHASE     +$31,200/yr
  Sectra PACS                    REPURCHASE     +$322,400/yr
  Philips IntelliSpace           REPURCHASE     +$169,000/yr
  MUSE                           REPURCHASE     +$20,800/yr
  Lumedx Apollo / CardioPACS     REPURCHASE     +$9,100/yr
  Acuo VNA                       REPURCHASE     +$31,200/yr
  Philips Digital Pathology      REPURCHASE     +$10,400/yr
  RapidAI                        REPURCHASE     +$9,100/yr
  EchoPACS                       REHOST         +$1,300/yr
  Hyland OnBase / VNA            REPURCHASE     +$144,300/yr
  CardioAccess                   REHOST         +$1,300/yr
  DynaCad Radiology              REHOST         +$1,300/yr
  Medicom                        REHOST         +$1,300/yr
  Infinitt Dental PACS           REPURCHASE     +$31,200/yr
  ATNA Audit Repository          REHOST         +$18,200/yr
  Butterfly Link                 REPURCHASE     +$9,100/yr
  HeartFlow Connect              REPURCHASE     +$10,400/yr
  Vizai                          REPURCHASE     +$2,600/yr
  GE Unified Viewer              REHOST         +$22,100/yr
  LifeImage / Expedo             REPURCHASE     +$10,400/yr
  Magview                        REPURCHASE     +$11,700/yr
  PenRad                         REPURCHASE     +$2,600/yr
  Circle                         REPURCHASE     +$1,300/yr
  Nuance PowerShare              REPURCHASE     +$23,400/yr
  Nuance PowerScribe One         REPURCHASE     +$1,300/yr
  Optellum / Lung AI             REPURCHASE     +$2,600/yr
  cNeuro                         REHOST         +$1,300/yr
  Volpara                        REPURCHASE     +$1,300/yr
  ACRConnect                     REPURCHASE     +$1,300/yr
  Teamplay                       REPURCHASE     +$1,300/yr
                                            ───────────
                                  Subtotal:  +$936,000/yr

Retiring apps (consolidating into successor systems):
                                            ───────────
                                  Subtotal:  +$0/yr

Structural RETAIN apps:                     $0 (out of migration scope)
  Philips Perinatal · Philips Patient Monitoring · PaceArt · Mosaiq · Hologic Apex

PORTFOLIO RUN-RATE ROI:                     +$1,432,900/yr
(achieved Year 2 onward, conservative baseline)

For each app's full cost breakdown, see the Provisioning Agent panel on its App Detail Agents tab.

How We Decide Migration Risk

Migration Risk is shown as HIGH, MEDIUM, or LOW on each app. The classification follows three simple rules evaluated in order:

HIGH
The app has hardware integration OR life-safety classification. These systems are operationally risky to move regardless of evidence quality — moving them is a structural problem, not a data problem.
MEDIUM
The app handles PHI AND we haven't confirmed the vendor's Business Associate Agreement. Compliance risk during transition that must be closed before migration.
LOW
Everything else. No hardware constraints, no unresolved PHI compliance gaps.

Risk is structural before it is compliance, and compliance before operational. This ordering reflects what most often breaks migrations in practice.

MUSC HEALTH IMAGING & RADIOLOGY — MIGRATION RISK WORKED EXAMPLE

Rule 1 check: hardware_dependency OR life_safety_classification?
              If yes → HIGH risk

  Philips Perinatal                   life-safety + hardware              → HIGH
  Philips Patient Monitoring          life-safety + hardware              → HIGH
  PaceArt                             life-safety + hardware              → HIGH
  Mosaiq                              life-safety + hardware              → HIGH
  Siemens Syngo / Syngovia            life-safety + hardware              → HIGH

Rule 2 check: phi_handled AND baa_confirmed=UNKNOWN?
              If yes → MEDIUM risk

  LifeImage / Expedo                  PHI + BAA unconfirmed               → MEDIUM
  Volpara                             PHI + BAA unconfirmed               → MEDIUM

Rule 3 default: everything else → LOW risk

  GE Centricity / CCW                 BAA confirmed or no PHI constraint  → LOW
  Sectra PACS                         BAA confirmed or no PHI constraint  → LOW
  Acuo VNA                            BAA confirmed or no PHI constraint  → LOW

Of 36 imaging/radiology apps, 23 are HIGH risk (hardware or life-safety constraints), 2 are MEDIUM risk (PHI handlers with unconfirmed BAAs), and 11 are LOW risk. The risk classification doesn't change the recommendation — it tells stakeholders which apps require the most operational care during execution.

For each app's risk classification, see the Migration Risk field on its App Detail Agents tab Dependency Agent panel.

Why You Can Trust This

R-ARF is built so every conclusion can be audited:

  • Every number has documented provenance — Sourced from registries, customer-provided from your data, computed from formulas, or default assumption clearly labeled.
  • Financial calculations use a Python engine, not an AI model — The Provisioning Agent panel on each App Detail page shows the engine's inputs and outputs.
  • Twenty-two validator rules across six agents check the math before publication — If any rule fails, the report does not publish. The validator rules enforce structural correctness — rule firing logic, provenance tagging, confidence ceilings, cap interactions. Each agent's rule pass/fail status is visible in the Validator Status panel on the Portfolio Detail tab.
  • Recommendations follow explicit rules visible in the Decision Path — There is no hidden logic. The rule that fired and why is shown on each App Detail Overview tab.
  • 69 default assumptions were used in this analysis — All defaults are documented in the Defaults Glossary on Portfolio Detail and can be replaced with client-provided data in future runs.

If you want to verify a specific recommendation, three sources tell the complete story:

  1. The Decision Path text on the app's App Detail Overview tab — shows the rule that fired and why.
  2. The Provisioning Agent panel on its Agents tab — shows the cost math step-by-step.
  3. The Backlog tab — shows what would need to change to lift the confidence score.

R-ARF v4.4.1 methodology. Governance framework unchanged since v3.1 (provenance model, evidence dimensions, confidence caps, validator rules).