AnythingLLM

AnythingLLM is a private AI assistant that runs entirely on your computer. The page says there are no accounts, no API keys and no token limits, so you can chat with your own documents, search the web, and run tasks without sending your data to a cloud service.

The desktop app supports document knowledge. Every document you add becomes part of your agent's knowledge base, and the page says no data leaves your device. It also includes web scraping and search, dynamic model selection, background jobs and custom agent skills.

The Meeting Assistant transcribes and summarises meetings on your computer. It produces action items, key decisions and full transcripts when the call ends. The page says no bot joins your meeting and no audio goes to cloud processing. It works with any meeting platform.

Magic Echo is a dictation feature. Your speech appears as text, is refined in real time, and the app can use what is on your screen to give more precise answers. The page says it runs fully on device.

Setup has four steps. You download a single file for macOS, Windows or Linux. You pick a model, or let AnythingLLM recommend one for your hardware. Then you chat with documents, do research and build workflows locally. A custom tool builder and a developer API let you adapt it to your needs.

AnythingLLM Mobile for Android runs small models on the device, with web search, document chat, file generation, scheduled jobs and memory. You can pair it with the desktop app to reach larger models and your full workspaces.

The project is MIT licensed and open source. The page lists over 67k GitHub stars and more than 200 contributors. You can self-host a multi-user deployment on your own servers or use the desktop app alone.

The page says the desktop app is free. It does not list paid plans, so this listing does not state a price beyond that.