How hrishikesh-prasad-r/info-vac works
InfoVac is a competitive-intelligence platform for researching loyalty programs. Analysts submit a program, the backend discovers and scrapes sources, extracts and verifies structured fields, then saves results and a brief for the workspace; users can also ask cited questions, compare programs, and review program evolution. The graph preserves the documented pipeline and major UI capabilities. Some connections are based on README-level architecture because the relevant implementation excerpts were not provided.
Main entry points: nodes.py Pipeline stages · main.py FastAPI endpoints · MultiFlowWorkspace.tsx Analyst views · page.tsx Analyst workspace
Components 31
User workspaces
- Analyst workspace
frontend/app/page.tsx - Analyst views
frontend/components/analyst/MultiFlowWorkspace.tsx - Chat widget
frontend/components/analyst/ChatWidget.tsx - Evidence drawer
frontend/components/analyst/EvidenceDrawer.tsx - PDF exporter
frontend/components/analyst/ExportBar.tsx - Admin dashboard
frontend/app/admin/page.tsx - Comparator picker
frontend/components/admin/ComparatorPicker.tsx
API and orchestration
- FastAPI endpoints
backend/main.py - Pipeline graph
orchestrator/graph.py - Pipeline stages
orchestrator/nodes.py - Progress events
orchestrator/events.py
Research pipeline
- Source discovery
backend/retriever.py - Source classifier
backend/classifier.py - Structured extraction
backend/extractor.py - Extraction schemas
backend/extraction_schemas.py - Verification gate
backend/gate.py - Confidence scoring
backend/verifier.py
Analysis services
- Executive brief
backend/narrator.py - Program comparison
backend/comparator.py - Program evolution
backend/routers/evolution.py - Cited Q&A
backend/chat.py - LLM routing
backend/llm_client.py - Key failover
backend/key_broker.py
State and search
- PostgreSQL
- Text embeddings
backend/embeddings.py - Qdrant vector store
backend/qdrant_client.py - RAG retrieval
backend/retriever.py - Domain records
backend/models.py - Database sessions
backend/db.py
Other components
- Analyst
- Admin
Connections 36
- Analyst → Analyst workspace: uses
README.md - Analyst workspace → Analyst views: renders
frontend/app/page.tsx - Analyst views → Chat widget: renders
frontend/components/analyst/MultiFlowWorkspace.tsx - Analyst views → Evidence drawer: renders
frontend/components/analyst/MultiFlowWorkspace.tsx - Analyst views → PDF exporter: renders
frontend/components/analyst/MultiFlowWorkspace.tsx - Admin → Admin dashboard: uses
README.md - Admin dashboard → Comparator picker: renders
frontend/components/admin/ComparatorPicker.tsx - Analyst workspace → FastAPI endpoints: requests research
README.md - Admin dashboard → FastAPI endpoints: requests analytics
README.md - FastAPI endpoints → Pipeline graph: starts pipeline
backend/main.py - Pipeline graph → Pipeline stages: dispatches stages
orchestrator/graph.py - Pipeline stages → Source discovery: discovers sources
orchestrator/nodes.py - Source discovery → Source classifier: classifies sources
backend/retriever.py - Pipeline stages → Structured extraction: extracts fields
orchestrator/nodes.py - Structured extraction → Extraction schemas: validates output
backend/extractor.py - Pipeline stages → Verification gate: checks evidence
orchestrator/nodes.py - Pipeline stages → Confidence scoring: scores confidence
orchestrator/nodes.py - Pipeline stages → Executive brief: generates brief
orchestrator/nodes.py - Structured extraction → LLM routing: requests completion
backend/extractor.py - LLM routing → Key failover: rotates keys
backend/llm_client.py - Pipeline stages → Progress events: emits progress
orchestrator/nodes.py - Progress events → PostgreSQL: writes events
orchestrator/events.py - FastAPI endpoints → PostgreSQL: reads and writes
backend/main.py - Pipeline stages → Text embeddings: chunks and embeds
orchestrator/nodes.py - Pipeline stages → Qdrant vector store: stores vectors
orchestrator/nodes.py - Text embeddings → Qdrant vector store: creates vectors
README.md - Chat widget → Cited Q&A: asks questions
README.md - Cited Q&A → RAG retrieval: retrieves context
backend/chat.py - RAG retrieval → Qdrant vector store: searches vectors
- Comparator picker → FastAPI endpoints: requests comparison
- FastAPI endpoints → Program comparison: routes comparison
backend/main.py - FastAPI endpoints → Program evolution: routes evolution
backend/main.py - Program evolution → LLM routing: requests analysis
backend/routers/evolution.py - FastAPI endpoints → Domain records: uses records
backend/main.py - FastAPI endpoints → Database sessions: opens sessions
backend/main.py - Database sessions → PostgreSQL: connects
backend/db.py