Use Case 1
Contract Design from Metric Selection to Configuration
Trigger: New BH VBC contract opportunity or contract renewal
1.Stage 1: select candidate quality metrics with GREEN/YELLOW/RED feasibility flags
2.Route metric slate to clinical leadership for sign-off (HITL)
3.Stage 2: run 24-month baseline measurement split W1/W2
4.Stage 3: produce contract configuration analysis (HITL—agent does not commit terms)
5.Cross-reference network adequacy for contracted geographies
6.Generate executive package with three shared-savings scenarios
✓
Leadership negotiates from a baseline-validated, scenario-modeled, network-checked starting position.
Use Case 2
Continuous Performance Tracking & Provider Scorecards
Trigger: Active VBC contract performance period
1.Refresh quality metrics monthly using claims_runout aggregates
2.Apply CMS HCC risk adjustment to financial measures
3.Generate provider scorecards with composite scores
4.Sort bottom quartile to Stage 8 targeting queue
5.Produce quarterly Stage 6 reports for medical_director sign-off
6.Aggregate to PHI-free payer-facing submission after sign-off
✓
Performance gaps surface monthly with provider-level detail; reporting goes external only after medical_director attestation.
Use Case 3
Renewal Evidence Package
Trigger: Contract approaching renewal window
1.Aggregate full-period quality and financial outcomes
2.Compare achieved savings to projection accuracy band
3.Build renewal scenarios with updated benchmarks
4.Cross-reference network adequacy for any geographic changes
5.Assemble renewal evidence package with CMS-compliant aggregations
6.Route to leadership (Stage 9 is HITL-only—agent does not commit terms)
✓
Renewals negotiated with full performance evidence and three-scenario shared-savings projections, no autonomous commitments.
Integrations
Claims Run-Out Warehouse
CMS HCC Risk Adjustment Tables
HEDIS Measure Library (NCQA)
Network Adequacy Database
Executive BI / Reporting
Approval Workflow (RequestApproval)
Industries
healthcare
See It In Action
A sample conversation with this agent
Stage 1: pick quality metrics for the new BH VBC contract with the Texas IPA.
Pulling candidate measure library and applying feasibility flags. For each candidate I'm checking data coverage, denominator size (N≥30 for GREEN), and 24-month baseline availability split into W1 and W2 windows. Top 8 candidates with GREEN flags: Antidepressant Medication Management (continuation + acute phases), Follow-Up After ED Visit for Mental Illness (7 and 30 day), Initiation and Engagement of SUD Treatment, Depression Remission at 12 Months, Adherence to Antipsychotic Medications. Three more come back YELLOW (denominator near threshold), four RED (data gaps). Stage 2 cannot run without explicit clinical leadership sign-off on Stage 1—routing the candidate slate now.
Show me the shared-savings scenarios for the renewal of the Magellan ACO contract.
Pulling claims_runout aggregates—paid_amount_pmpm, benchmark_pmpm, variance_vs_benchmark, and completion_factor. Not recomputing TCoC from claim lines (single source of truth rule). Applying completion factor to project final-year savings: $4.2M gross savings before quality gate. Three scenarios: (1) Full quality gate met, 1.0x multiplier = $4.2M shareable; (2) Partial gate, 0.85x = $3.57M; (3) Gate not met, 0.0x = no shareable savings. Leadership picks the scenario—I do not commit to terms. Cross-referenced against pbo6no_list_adequacy_calculations for CO, TX, MA—network supports proposed thresholds.
Generate the bottom-quartile provider list for Stage 8 targeting.
Provider scorecards sorted by composite score descending. Bottom quartile (62 providers) auto-feeds the Stage 8 targeting queue. By specialty: 38 outpatient BH providers, 14 IOP/PHP programs, 10 inpatient psych. Top contributing factors to low scores: AMM continuation phase below 45% (24 providers), FUH-7 day below 35% (19 providers), SUD initiation below 28% (8 providers). I do not design the care-model response—Stage 8 is HITL-only, the memo stops at cohort and provider targeting and clinical leadership designs the intervention.