Mintplex-Labs/anything-llm

What it solves

AnythingLLM is an all-in-one AI application designed to let users create a private, fully-featured ChatGPT-like experience. It removes the friction of setting up complex AI stacks by providing built-in vector databases, document pipelines, and agent capabilities out of the box, allowing users to chat with their own documents privately and securely.

How it works

The system operates as a monorepo with a React frontend for content management, a NodeJS Express server for LLM and vector database interactions, and a dedicated document collector for parsing files. It allows users to connect to various local or cloud-based LLM providers, embedder models, and vector databases, providing a hyper-configurable environment where documents are ingested and indexed for retrieval-augmented generation (RAG).

Who it’s for

It is designed for individuals and organizations who want a private AI instance that can handle multi-user support, custom AI agents, and document-based chatting without needing extensive technical configuration or manual infrastructure setup.

Highlights

• Dynamic Model Routing: Automatically routes chats to the best provider and model based on user-defined rules.
• No-code AI Agent Builder: Create custom agents to automate complex workflows and browse the web.
• Extensive Compatibility: Supports a vast array of LLMs (OpenAI, Anthropic, Ollama, etc.), embedders, and vector databases (LanceDB, Pinecone, Chroma, etc.).
• Multi-user Support: Includes permissioning and access control for multi-user instances (Docker version).
• Enterprise-ready: Features a custom embeddable chat widget for websites and a full developer API for integrations.