AnythingLLM

AnythingLLM is an AI application for working with documents and AI agents in one interface. It is described as local and private by default, with desktop use centered on running AI tasks without requiring setup or code.

The product supports chatting with documents, using AI agents, and working with custom models. It can run a preferred LLM locally, connect to local or cloud LLM engines, or use enterprise models from OpenAI, Azure, AWS, and more. It also supports text-only and multi-modal LLMs, with images or audio mentioned as usable in the same interface. Document support includes PDFs, Word documents, CSV files, codebases, and importing documents from online locations. The site also says it includes built-in defaults for the LLM, embedder, vector database, storage, and agents, and that nothing is shared unless allowed.

AnythingLLM Desktop is available for MacOS, Windows, and Linux, and is presented as a one-click install. The desktop version is not SaaS and does not require a signup to use the full suite of tools locally. The same page also mentions hosted and self-hosted options for team use, with multi-user access, full isolation between tenants, and admin control over what users can do and see. It is open source, free to use, and MIT licensed. The page further describes a built-in developer API, plus a growing ecosystem of plugins and integrations, custom agents, data connectors, and data loaders.