Overview
RAGFlow combines a vector store, ElasticSearch, DeepDoc segmentation and GraphRAG reasoning to deliver truthful, citation-backed answers from heterogeneous data.
Key Capabilities
- Multi-format parsing (PDF, Office, code, web, DB dumps)
- DeepDoc semantic segmentation for high-recall retrieval
- GraphRAG pipeline with entity linking & reranking
- Code-sandbox component for live code execution in answers
- Docker / Helm deploy, Python SDK & MCP server integrations