eDiscovery Investigation Copilot
Outcome
Help the litigation team work a document population custodian by custodian: cull with defensible search terms, prioritize responsive material, surface privilege risk before production, and keep a chain-of-custody and review-decision record that survives a motion to compel.
| Portfolio attribute | Value |
|---|---|
| Primary persona | Litigation team |
| Primary aggregate | CustodianReviewSet |
| Cosmos DB containers | 1 |
| Grounding data | email, chat, documents, metadata, issue tags |
| 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 | Investigation rework | 8-18% reduction | (baseline reopened or corrected investigation-task rate - assisted rate) / baseline rate * 100 |
| Quality or risk | Evidence discovery time | 25-45% reduction | (baseline median evidence-search minutes - assisted median evidence-search minutes) / baseline median evidence-search minutes * 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: Email, chat, documents, and metadata create a retrieval-heavy workflow where faster discovery can reduce repeated investigation work.
Pilot guardrail: Preserve privilege, chain of custody, access controls, and explicit handling of weak or incomplete 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 investigation 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. Custodian collection and chain of custody | Forensic collection service | CollectedItem → custodianPopulation | Policy controlled |
| 2. Search term culling and sampling | eDiscovery investigation copilot | SearchTermResult → culledReviewSet | Human required |
| 3. Responsiveness prioritization | eDiscovery investigation copilot | culledReviewSet → prioritizedQueue | Policy controlled |
| 4. Privilege screening and family propagation | Privilege screening service | PrivilegeIndicator → privilegeQueue | Policy controlled |
| 5. Attorney coding and production certification | Reviewing attorney and certifying counsel | ReviewDecision → ProductionBatch | Human required |
Technical architecture

Cosmos DB data model
| Container | Hierarchical partition key | Documents | Search paths |
|---|---|---|---|
custodianReviewSets | /tenantId, /custodianId | Custodian, CollectedItem, SearchTermResult, ReviewDecision, PrivilegeIndicator, ProductionBatch | /itemVectorEmbedding, /itemVector |
Query contract
| Query | Container | Scope | Purpose |
|---|---|---|---|
| UnreviewedPrioritizedItems | custodianReviewSets | single-partition | Serve the reviewer the highest-value uncoded items for this custodian. |
| ConceptSearchAcrossCustodian | custodianReviewSets | single-partition | Run a concept query over the custodian population to find responsive material that keyword terms missed. |
| PrivilegeQueueByConfidence | custodianReviewSets | single-partition | Queue items with privilege indicators for attorney review before any production. |
| InconsistentFamilyCoding | custodianReviewSets | single-partition | Detect families where an attachment and its parent received conflicting responsiveness or privilege calls. |
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/legal-services/ediscovery-investigation-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