Fortemi/fortemi
📦 Open Source ProjectFortemi
A high-performance, self-hosted AI knowledge base featuring hybrid search, knowledge graphs, and native MCP server support.
Fortemi leverages the memory safety and performance of Rust combined with the reliability of PostgreSQL to create a powerful AI-native knowledge management system. At its core, the platform implements advanced retrieval techniques, utilizing pgvector for semantic search, Full-Text Search (FTS), and Reciprocal Rank Fusion (RRF) to ensure high-accuracy information retrieval.
Beyond simple RAG, Fortemi constructs a knowledge graph to map relationships between data points, enhancing contextual understanding. It supports diverse data ingestion pipelines, including vision processing and audio transcription with speaker diarization. The architecture is built to be modular, offering multi-provider LLM inference support (Ollama, OpenAI, OpenRouter, llama.cpp) and acting as an MCP server, allowing it to serve as a centralized intelligence hub for AI agents and local applications.
💡Highlights
- ├─Hybrid search: pgvector, FTS, RRF
- ├─Native MCP server integration
- └─Multimodal: vision, audio, diarization
🎯For
- ├─AI Engineers
- ├─Knowledge Management Enthusiasts
- └─Self-hosting Advocates