ragflow — RAGFlow is an open-source Retrieval-Augmented Generation (RAG) engine that combines RAG with Agent capabilities to provide a high-fidelity context layer for Large Language Models.

RAGFlow is an open-source Retrieval-Augmented Generation (RAG) engine that combines RAG with Agent capabilities to provide a high-fidelity context layer for Large Language Models.

About this project

RAGFlow is an open-source RAG engine designed to transform complex, unstructured data into production-ready AI systems. It focuses on deep document understanding and provides a streamlined workflow for enterprises to integrate LLMs with their own data. Key capabilities include: - Deep document understanding for knowledge extraction from unstructured data with complex formats. - Template-based chunking that is intelligent and explainable. - Support for diverse data sources including Word, Slides, Excel, TXT, images, scanned copies, and web pages. - Grounded citations to reduce hallucinations by providing traceable references to text chunks. - An automated RAG orchestration workflow with configurable LLMs, embedding models, and fused re-ranking. - Integration with Agentic workflows, MCP, and a Python/JavaScript code executor. - Data synchronization from external sources such as Confluence, S3, Notion, Discord, and Google Drive. The system can be self-hosted via Docker and supports various document engines, including Elasticsearch (default) and Infinity.