Atomic: Open-source AI knowledge base with semantic search and wiki synthesis
Atomic is a new open-source, self-hosted personal knowledge base that uses AI to automatically organize notes, articles, and web clips into a knowledge graph. The app, available as a desktop app via Tauri, a headless server, an iOS app, and a browser extension, features vector embeddings for semantic search, LLM-generated wiki articles with inline citations, and an AI chat agent that searches the user's notes. It also offers a force-directed graph visualization and an MCP server for integration with tools like Claude and Cursor. Atomic automatically tags and links atoms (individual pieces of knowledge) and provides a daily AI-generated briefing. The project is hosted at atomicapp.ai.
Key facts
- Atomic is a local-first, AI-augmented personal knowledge base.
- It is open source and self-hosted.
- Available as a desktop app via Tauri, headless server, iOS app, and browser extension.
- Uses vector embeddings for semantic search across notes, articles, and web clips.
- LLM-generated wiki articles synthesize content under tags with inline citations.
- Includes an AI chat agent that searches notes mid-conversation with source citations.
- Visualizes knowledge as a force-directed graph.
- Provides an MCP server for integration with Claude, Cursor, and other MCP clients.
Entities
Institutions
- Atomic
- Claude
- Cursor
- MCP