OpenRAG

OpenRAG is a self-hosted platform for searching documents and asking questions about them through an AI chat interface. It brings Langflow, OpenSearch, and Docling together for developers and teams that want a document knowledge base with control over its retrieval workflows. It's open source under Apache 2.0 and supports self-managed deployment with Docker or Podman.

The core services run in containers. Docling processes documents locally, including PDFs that need advanced parsing, while OpenSearch stores documents and their embeddings for semantic and hybrid search. Documents can come from your filesystem or cloud storage through OAuth-authorized connectors. Those connectors access external services; the platform also integrates with LLM providers, so self-hosting doesn't by itself mean every part runs offline.

Langflow's embedded visual editor lets you change the supplied workflows or build your own. You can replace the language model or other components, add custom components, connect MCP servers, and call APIs. Agentic retrieval workflows support re-ranking and coordination between multiple agents, giving teams room to adjust how the system finds material and prepares answers.

A built-in MCP server makes the knowledge base accessible to assistants such as Cursor, Claude Desktop, and IBM Bob. It provides document ingestion, semantic search, and chat using retrieved knowledge. Python and TypeScript/JavaScript SDKs let developers add these capabilities to their own applications.