How medtronic-labs/data-insights-copilot works
Data Insights AI-Copilot is an enterprise data-analysis application: users configure agents and data sources, then ask questions through a React chat interface. The documented query flow routes intent to structured SQL, vector retrieval, or a hybrid response, with results rendered as answers and charts. The repository tree and excerpts support the API, chat, data-source, embedding, and administration modules; several pipeline connections and background-job details are documented in the README rather than shown in sampled implementation code.
Main entry points: chat/service.py Chat orchestration · app.py FastAPI application · sql_service.py SQL analysis · data_sources/service.py Data source management
Components 23
User experience
- React application
frontend/src/App.tsx - Chat interface
frontend/src/pages/ChatPage.tsx
Question answering
- Chat routes
backend/app/modules/chat/routes.py - Chat orchestration
backend/app/modules/chat/service.py - Intent routing
backend/app/modules/chat/intent_classifier.py - SQL analysis
backend/app/modules/chat/sql_service.py - Hybrid synthesis
backend/app/modules/chat/comparison_engine.py
Data and retrieval
- Vector retrieval
backend/app/modules/embeddings/retrieve.py - Data source management
backend/app/modules/data_sources/service.py - Embedding service
backend/app/modules/embeddings/service.py
Administration
- Agent configuration
backend/app/modules/agents/service.py - AI model registry
backend/app/modules/ai_models/service.py - User management
backend/app/modules/users/service.py
Platform services
- FastAPI application
backend/app/app.py - Language models
backend/app/core/llm/factory.py - Authentication and RBAC
backend/app/core/auth/permissions.py - Audit logging
backend/app/modules/audit/service.py - Observability
backend/app/modules/observability/analytics_service.py - RAG evaluation
eval/pipeline/e2e_evaluator.py
Other components
- Analyst
- Clinical database
- Vector store
- Internal configuration store
Connections 24
- Analyst → React application: uses
- React application → Chat interface: opens chat
frontend/src/App.tsx - Chat interface → FastAPI application: sends questions
- FastAPI application → Chat routes: routes requests
backend/app/app.py - Chat routes → Chat orchestration: calls
- Chat orchestration → Intent routing: classifies intent
- Chat orchestration → SQL analysis: runs SQL
- Chat orchestration → Vector retrieval: retrieves context
- Chat orchestration → Hybrid synthesis: synthesizes results
- SQL analysis → Clinical database: queries
- Vector retrieval → Vector store: searches
- SQL analysis → Language models: generates SQL
- Hybrid synthesis → Language models: generates answer
- Data source management → Clinical database: connects runtime
- Data source management → Embedding service: indexes documents
- Embedding service → Vector store: stores vectors
- Agent configuration → Internal configuration store: persists configuration
- AI model registry → Internal configuration store: persists registry
- Authentication and RBAC → User management: provisions identities
backend/app/core/auth/permissions.py - User management → Internal configuration store: stores users
- FastAPI application → Authentication and RBAC: enforces access
- FastAPI application → Audit logging: records actions
- FastAPI application → Observability: reports analytics
- RAG evaluation → Chat orchestration: evaluates pipeline