You want to ask questions of material kept on your own computer without touching a command line. AnythingLLM ships as a desktop app: install it, create workspaces, import documents and use a local or remote model service as the generation backend, with chat history stored on the machine. Moving to a new computer means migrating workspaces yourself. Supported formats and retrieval quality depend on your backend; check the project repository.
What it does
- A desktop app, no command line required
- Documents managed by workspace keep Q and A scoped
- Built-in local vector storage keeps material on your machine
- Local and remote model services both work
- Chat history can be saved and organized
Who it is for
- Material is private or internal and should not go to another service
- A document assistant is wanted on your own computer
- A local model is already in use but lacks a GUI
- Separate projects need separate corpora to avoid mixing
How to get started
- Create one workspace with a small batch of documents to test retrieval
- Use a local model first; consider remote services only if needed
- Create a new workspace for a different document set to compare answer quality
