Impact Reporting Copilot
Outcome
Let the program director produce an impact report that ties indicator results to the program's logic model, reconciles outcome data with financial spend, and pairs quantitative results with consented participant stories, without overstating attribution.
| Portfolio attribute | Value |
|---|---|
| Primary persona | Program director |
| Primary aggregate | ProgramImpactPeriod |
| Cosmos DB containers | 2 |
| Grounding data | outcomes, surveys, financials, stories, indicators |
| 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 | Reporting rework | 6-14% reduction | (baseline corrected or repeated reporting-task rate - assisted rate) / baseline rate * 100 |
| Quality or risk | Evidence discovery time | 15-30% reduction | (baseline median evidence-search minutes - assisted median evidence-search minutes) / baseline median evidence-search minutes * 100 |
| Business outcome | Approved funder-report turnaround | 5-10% reduction | (baseline median report days - assisted median report days) / baseline median report days * 100 |
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, survey, partner, and finance evidence can reduce report corrections and modestly shorten approval turnaround.
Pilot guardrail: Preserve measurement caveats, attribution limits, participant-story consent, and source evidence.
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 reporting rework and evidence discovery time 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. Indicator data consolidation and quality rating | Data connectors with data steward | IndicatorResult → ratedIndicatorSet | Policy controlled |
| 2. Financial reconciliation to general ledger | Finance analyst | FinancialAllocation → reconciledSpend | Human required |
| 3. Story selection under consent scope | Impact reporting copilot | ParticipantStory → eligibleStorySet | Policy controlled |
| 4. Report synthesis with attribution caveats | Impact reporting copilot | EvaluationChunk → ImpactReportDraft | Policy controlled |
| 5. Program director and funder-variant approval | Program director | ImpactReportDraft → approvedImpactReport | Human required |
Technical architecture

Cosmos DB data model
| Container | Hierarchical partition key | Documents | Search paths |
|---|---|---|---|
programImpactPeriods | /tenantId, /programId | Program, IndicatorResult, FinancialAllocation, ParticipantStory, ImpactReportDraft | /storyVectorEmbedding, /storyVector |
evaluationLiteratureArchive | /tenantId, /evalDocId | EvaluationChunk | /chunkVectorEmbedding, /chunkVector |
Query contract
| Query | Container | Scope | Purpose |
|---|---|---|---|
| IndicatorPerformanceVersusTarget | programImpactPeriods | single-partition | Compare actuals to targets across the logic model for the reporting period. |
| RestrictedFundSpendReconciliation | programImpactPeriods | single-partition | Report spend by funding source with restriction class for funder-specific variants. |
| ConsentEligibleStories | programImpactPeriods | single-partition | Select only stories whose consent scope permits the intended publication channel. |
| AttributionCaveatRetrieval | evaluationLiteratureArchive | cross-partition-bounded | Retrieve evaluation methodology text that bounds what the program may claim to have caused. Intentional cross-partition fan-out across the evaluationLiteratureArchive reference container, bounded by the tenant prefix, the comparison-group flag and a publication lookback window, returning only the top-ranked evaluation chunks. |
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/impact-reporting-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