// the find
GerevAI/gerev
🧠 AI-powered enterprise search engine 🔎
Gerev is a self-hosted enterprise search tool that indexes Slack, Confluence, Jira, Google Drive, and a handful of other workplace data sources, then lets you query them with natural language. It uses FAISS for vector search combined with BM25 for hybrid retrieval. Aimed at small-to-medium engineering teams who want a single search box across their internal tools.
The plugin architecture in `app/data_source/api/base_data_source.py` with `dynamic_loader.py` is genuinely well-designed — adding a new data source is mechanical, not a surgery. The hybrid search approach (FAISS + BM25 in `app/indexing/`) is a reasonable choice that handles both semantic and keyword queries without requiring a heavyweight search backend. Docker-first deployment with volume persistence means getting it running is actually a one-liner for most people. Community contributed 6+ integrations, which is a real signal that the extension surface works.
Last commit was December 2023 — this project is effectively dead. The 'In PR' integrations for Zendesk, Notion, Trello, Teams have been sitting there for over two years with no merge. Requires NVIDIA GPU for the self-hosted LLM path; the CPU fallback exists but the README buries it and performance will be rough. No authentication in the community version at all — whoever can reach port 80 can search your internal docs, which is a hard no for most organizations. The cloud offering redirects to a third-party SaaS (klu.so) that has nothing to do with this repo, so the project's commercial future is unclear at best.