RigRAG

RigRAG is a RAG-as-a-Service project for AI Agents. The platform allows users to upload documents, automatically process them, and provide them as searchable memory spaces via REST API or MCP.

The system combines semantic vector search with BM25 full-text search and Reciprocal Rank Fusion. Optional neural reranking further improves the relevance of search results. This enables AI Agents to access external knowledge in a targeted way, without having to build a custom retrieval pipeline for every project.

The focus was on a production-ready, secure, and developer-friendly architecture with isolated knowledge spaces, encrypted API keys, user-based data separation, and a credit-based cost model.