RAGFlow

RAG engine built on deep document understanding with grounded cites.

RAGFlow is an open-source RAG engine built around deep document understanding, with a focus on parsing complex, heterogeneous documents (PDFs, tables, images, scans) into structured, citation-grounded chunks. It provides templated chunking, an end-to-end pipeline from ingestion to grounded question answering with traceable references, and a web UI for managing knowledge bases. It is notable for emphasizing layout-aware document parsing to reduce hallucination in answers.