Caseworker Program Copilot
Outcome
Help the caseworker move a participant through a multi-program service plan by grounding eligibility, benefit interactions and referral options in current program rules and the participant's own consented record, so scarce caseworker time goes to the relationship rather than to rule lookup.
| Portfolio attribute | Value |
|---|---|
| Primary persona | Caseworker |
| Primary aggregate | ParticipantServiceCase |
| Cosmos DB containers | 2 |
| Grounding data | case notes, program rules, services, eligibility |
| Development estimate | USD 60-220 monthly dev estimate |
| Runtime options | Python FastAPI and C# ASP.NET Core |
Productivity hypothesis
These measures reflect the workflow, evidence, decision boundary, and industry risk for this use case.
| Impact type | Measure | Conservative internal assumption | Customer measurement |
|---|---|---|---|
| Productivity | Rules and participant-context discovery time | 15-25% reduction | (baseline median evidence-search minutes - assisted median evidence-search minutes) / baseline median evidence-search minutes * 100 |
| Quality or risk | First-pass acceptance | 5-12 percentage-point improvement | assisted percentage accepted without material rework - baseline percentage accepted without material rework |
Important: Conservative internal pilot assumptions only; not public benchmarks, commitments, or expected customer results. Validate against a comparable baseline or matched control cohort with mandatory human review.
Rationale: Program rules, eligibility, services, and participant context support faster preparation of reviewable case plans.
Pilot guardrail: Respect participant consent and require human approval for service plans, referrals, and eligibility decisions.
Customer validation plan
- Baseline the current workflow with representative users and tasks.
- Run the accelerator on the same task mix with human review retained.
- Measure rules and participant-context discovery time and first-pass acceptance with the documented formulas.
- Replace internal assumptions with customer-observed results before a business case.
Why Azure Cosmos DB
This AI solution needs an operational data plane, not only a vector index:
- Hybrid vector and full-text retrieval combines semantic recall with exact identifiers, codes, names, and domain language.
- Hierarchical partition keys and metadata predicates keep tenant and domain boundaries in the normal query path.
- Operational JSON, source chunks, workflow and agent state, feedback, and audit records remain together for low-latency reads and governed writes.
- Change feed reacts to source and operational updates without polling.
- Autoscale throughput, configurable consistency, and multi-region distribution support variable demand and enterprise availability.
Domain workflow

| Stage | Actor | Input → output | Approval |
|---|---|---|---|
| 1. Strengths and needs intake with consent | Caseworker with participant | participantConversation → ParticipantCase | Human required |
| 2. Program rule and eligibility grounding | Caseworker program copilot | ParticipantCase → ProgramRuleChunk | Policy controlled |
| 3. Benefit interaction analysis | Caseworker program copilot | ProgramRuleChunk → BenefitInteractionFinding | Policy controlled |
| 4. Service plan co-creation | Caseworker with participant | BenefitInteractionFinding → ServicePlanGoal | Human required |
| 5. Referral with consent and capacity check | Caseworker | ServicePlanGoal → ReferralRecord | Human required |
Technical architecture

Cosmos DB data model
| Container | Hierarchical partition key | Documents | Search paths |
|---|---|---|---|
participantServiceCases | /tenantId, /caseId | ParticipantCase, ServicePlanGoal, ReferralRecord, EncounterNote, BenefitInteractionFinding | Operational fields only |
programRuleLibrary | /tenantId, /programCode | ProgramRuleChunk | /chunkVectorEmbedding, /chunkVector |
Query contract
| Query | Container | Scope | Purpose |
|---|---|---|---|
| CaseloadByNextAction | participantServiceCases | cross-partition-bounded | Order the caseworker's caseload by the goals and referrals with the nearest due dates. Intentional cross-partition caseworker worklist: bounded by the tenantId prefix, the responsible caseworker and a due-date horizon. |
| EffectiveProgramRuleRetrieval | programRuleLibrary | cross-partition-bounded | Retrieve only the program and funder rules in force for the participant's enrollment date. |
| OpenBenefitInteractions | participantServiceCases | single-partition | Surface unresolved findings where one enrollment would harm another benefit. |
| ReferralOutcomeHistory | participantServiceCases | single-partition | Review which partner referrals actually closed successfully to inform the next handoff. |
Python and C# services route each document type to its declared container and hierarchical partition key. ETag-based writes protect mutable aggregates, while append-only records and bounded cross-partition queries preserve scale.
Azure deployment

Clone the source repository, authenticate with Azure, and run:
Set-Location accelerators/nonprofit-social-impact/caseworker-program-copilot
az login
./Deploy.ps1 -Runtime python -Location eastus2
Change python to csharp to deploy the .NET implementation.
Explore the implementation
- Launch signed-in demo
- Implementation article
- Domain data model
- Workflow and API contract
- Python application
- C# application
- Domain container Bicep
- Platform Bicep
- Cost and configuration
- Source repository