How bharathmrr/agentic-graph-rag-as-a-service works
This is an Agentic Graph RAG platform: users upload documents, which are OCR-processed and transformed into ontology entities, deduplicated, embedded, and stored in graph and vector databases for visualization and retrieval-backed chat. The graph preserves the README’s major product branches, including multi-LLM support, query/ reasoning, and monitoring. Relationships are limited to calls shown in excerpts or workflow links documented in the README; several implementations are unsampled, so their internal connections are not inferred.
Main entry points: pipeline_runner.py Ingestion Orchestrator · file_processing_pipeline.py File Processing · enhanced_agentic_retrieval.py Agentic Retrieval · ApplicationChatBot.jsx AI Chat Interface
Components 21
User Experience
- Application UI
frontend/src/App.jsx - Document Upload
frontend/src/components/EnhancedDocumentUpload.jsx - Graph Visualization
frontend/src/components/EnhancedKnowledgeGraph.jsx - AI Chat Interface
frontend/src/components/ApplicationChatBot.jsx
Document Ingestion
- API Application
src/api/main.py - Processing Routes
backend/routes/enhanced_processing_routes.py - File Processing
backend/services/file_processing_pipeline.py - OCR Extraction
backend/utils/ocr_extractor.py - LLM Service
backend/services/enhanced_llm_service.py
Knowledge Construction
- Ontology Generation
src/ingestion/ontology_generator.py - Entity Resolution
src/ingestion/entity_resolution.py - Embedding Generation
src/ingestion/embedding_generator.py - Graph Construction
src/ingestion/graph_constructor.py - Ingestion Orchestrator
src/ingestion/pipeline_runner.py - Neo4j Graph Store
- ChromaDB Vector Store
Retrieval and Chat
- Retrieval Routes
src/api/routes/retrieval_routes.py - Agentic Retrieval
src/retrieval/enhanced_agentic_retrieval.py - Reasoning Stream
src/retrieval/enhanced_reasoning_stream.py
Other components
- User
- LLM Providers
Connections 24
- User → Application UI: uses
README.md - Application UI → Document Upload: opens upload
frontend/src/App.jsx - Document Upload → Processing Routes: submits documents
README.md - Processing Routes → File Processing: starts processing
backend/routes/enhanced_processing_routes.py - File Processing → OCR Extraction: extracts text
README.md - File Processing → LLM Service: requests AI processing
backend/services/file_processing_pipeline.py - LLM Service → LLM Providers: uses configured model
README.md - File Processing → Ontology Generation: generates ontology
README.md - Ingestion Orchestrator → Ontology Generation: generates ontology
src/ingestion/pipeline_runner.py - Ingestion Orchestrator → Entity Resolution: resolves entities
src/ingestion/pipeline_runner.py - Ingestion Orchestrator → Embedding Generation: generates embeddings
src/ingestion/pipeline_runner.py - Ingestion Orchestrator → Graph Construction: constructs graph
src/ingestion/pipeline_runner.py - Ontology Generation → Entity Resolution: deduplicates entities
README.md - Entity Resolution → Neo4j Graph Store: stores entities
README.md - Embedding Generation → ChromaDB Vector Store: stores vectors
README.md - Graph Construction → Neo4j Graph Store: writes graph
README.md - Neo4j Graph Store → Graph Visualization: supplies graph data
README.md - User → AI Chat Interface: asks questions
README.md - AI Chat Interface → Retrieval Routes: submits query
README.md - Retrieval Routes → Agentic Retrieval: routes retrieval
- Agentic Retrieval → ChromaDB Vector Store: searches vectors
README.md - Agentic Retrieval → Neo4j Graph Store: traverses graph
README.md - Agentic Retrieval → Reasoning Stream: forms response
- Reasoning Stream → AI Chat Interface: returns answer
README.md