Skip to main content
LEG-03 · Legal Services

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 attributeValue
Primary personaLitigation team
Primary aggregateCustodianReviewSet
Cosmos DB containers1
Grounding dataemail, chat, documents, metadata, issue tags
Development estimateUSD 60-220 monthly dev estimate
Runtime optionsPython FastAPI and C# ASP.NET Core

Productivity hypothesis​

These measures reflect the workflow, evidence, decision boundary, and industry risk for this use case.

Impact typeMeasureConservative internal assumptionCustomer measurement
ProductivityInvestigation rework8-18% reduction(baseline reopened or corrected investigation-task rate - assisted rate) / baseline rate * 100
Quality or riskEvidence discovery time25-45% 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​

  1. Baseline the current workflow with representative users and tasks.
  2. Run the accelerator on the same task mix with human review retained.
  3. Measure investigation rework and evidence discovery time with the documented formulas.
  4. 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​

Domain workflow

StageActorInput → outputApproval
1. Custodian collection and chain of custodyForensic collection serviceCollectedItem → custodianPopulationPolicy controlled
2. Search term culling and samplingeDiscovery investigation copilotSearchTermResult → culledReviewSetHuman required
3. Responsiveness prioritizationeDiscovery investigation copilotculledReviewSet → prioritizedQueuePolicy controlled
4. Privilege screening and family propagationPrivilege screening servicePrivilegeIndicator → privilegeQueuePolicy controlled
5. Attorney coding and production certificationReviewing attorney and certifying counselReviewDecision → ProductionBatchHuman required

Technical architecture​

Technical architecture

Cosmos DB data model​

ContainerHierarchical partition keyDocumentsSearch paths
custodianReviewSets/tenantId, /custodianIdCustodian, CollectedItem, SearchTermResult, ReviewDecision, PrivilegeIndicator, ProductionBatch/itemVectorEmbedding, /itemVector

Query contract​

QueryContainerScopePurpose
UnreviewedPrioritizedItemscustodianReviewSetssingle-partitionServe the reviewer the highest-value uncoded items for this custodian.
ConceptSearchAcrossCustodiancustodianReviewSetssingle-partitionRun a concept query over the custodian population to find responsive material that keyword terms missed.
PrivilegeQueueByConfidencecustodianReviewSetssingle-partitionQueue items with privilege indicators for attorney review before any production.
InconsistentFamilyCodingcustodianReviewSetssingle-partitionDetect 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​

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​

Return to all accelerators