How tanmay7264/industrial-knowledge-intelligence works
Industrial Brain is an organizational-memory platform for industrial operations. Engineers add operational documents, which are parsed, chunked, enriched with entities, embedded, and indexed in vector and graph stores; they then ask questions or investigate failures and receive evidence-backed answers and action plans. Confirmed feedback can preserve operational knowledge for later use. The README also describes asset, alert, compliance, insight, and knowledge-risk capabilities; unsampled implementations are represented by their existing route or library directories. The excerpts do not establish every subsystem’s internal wiring, so uncertain links are omitted.
Main entry points: pipeline.ts Ingestion pipeline · api Operations APIs · route.ts Chat API · retrieve.ts Graph retrieval
Components 28
Engineer experience
- Ask Memory
src/app/query/query-page.tsx - Document Intelligence
src/app/ingest/page.tsx - Command Center
src/app/command/page.tsx - Failure investigations
src/app/incidents/[id]/page.tsx - AI reasoning graph
src/app/graph/page.tsx
Document ingestion
- Ingestion API
src/app/api/ingest/route.ts - Ingestion pipeline
src/lib/ingest/pipeline.ts - Document loaders
src/lib/ingest/loaders/index.ts - Semantic chunking
src/lib/ingest/chunk.ts - Entity extraction
src/lib/ingest/entities.ts
AI reasoning
- Embedding service
src/lib/ai/embeddings.ts - Chat API
src/app/api/chat/route.ts - RAG routing
src/lib/rag/router.ts - Evidence retrieval
src/lib/rag/retrieve.ts - Answer confidence
src/lib/rag/answer.ts
Operational intelligence
- Root-cause analysis
src/lib/agents/rca.ts - Playbook generation
src/lib/agents/playbook.ts - Engineer feedback
src/app/api/feedback/route.ts - Operations APIs
src/app/api - Knowledge risk
src/lib/agents/knowledge-risk.ts
Knowledge stores
- Vector index
src/lib/ingest/qdrant-store.ts - Qdrant
src/lib/clients/qdrant.ts - Knowledge graph builder
src/lib/graph/build.ts - Graph retrieval
src/lib/graph/retrieve.ts - Neo4j graph
src/lib/clients/neo4j.ts - Redis cache
src/lib/clients/redis.ts
Other components
- Plant engineer
- LLM provider
Connections 35
- Plant engineer → Document Intelligence: uploads documents
- Document Intelligence → Ingestion API: submits files
- Ingestion API → Ingestion pipeline: ingests files
- Ingestion pipeline → Document loaders: loads documents
- Ingestion pipeline → Semantic chunking: chunks text
- Ingestion pipeline → Entity extraction: extracts entities
- Ingestion pipeline → Embedding service: embeds chunks
- Ingestion pipeline → Vector index: stores chunks
- Ingestion pipeline → Knowledge graph builder: builds graph
- Vector index → Qdrant: writes vectors
- Knowledge graph builder → Neo4j graph: writes graph
- Plant engineer → Ask Memory: asks questions
- Ask Memory → Chat API: sends query
- Chat API → Embedding service: embeds query
- Chat API → RAG routing: routes query
- Chat API → Answer confidence: computes confidence
- Chat API → LLM provider: generates answer
- Chat API → Redis cache: caches answers
- RAG routing → Evidence retrieval: retrieves evidence
- Evidence retrieval → Graph retrieval: uses graph search
- Graph retrieval → Neo4j graph: queries graph
- Plant engineer → Failure investigations: opens incident
- Failure investigations → Operations APIs: loads incident
- Operations APIs → Root-cause analysis: generates RCA
- Operations APIs → Graph retrieval: retrieves lessons
- Plant engineer → Playbook generation: requests playbook
- Playbook generation → LLM provider: generates steps
- Plant engineer → Engineer feedback: confirms steps
- Engineer feedback → Redis cache: stores feedback
- Plant engineer → Command Center: views plant status
- Command Center → Operations APIs: loads operations data
- Operations APIs → Knowledge risk: builds risk report
- Operations APIs → Graph retrieval: queries incidents
- Operations APIs → Qdrant: counts documents
- Plant engineer → AI reasoning graph: explores evidence