AnythingLLM

A free local AI assistant for Windows, macOS, Linux and Android, with document chat, agents and a Docker option for self-hosted teams.

Alternative to:

AnythingLLM is an AI assistant for people who want document chat and agents on their own hardware. Its desktop app can work with local models without an account or cloud API key.

Document chat: Add PDFs, text files and Word documents to workspaces, then ask questions with source citations.
Agents and memory: Build agents without code, add custom skills and MCP tools, and run scheduled tasks with workspace knowledge and persistent memories.
Model choice: Use Ollama, LM Studio, LocalAI or llama.cpp-compatible models, or connect providers such as OpenAI and Anthropic; routing rules can select models for different conversations.
On-device productivity: The desktop meeting assistant transcribes and summarizes calls locally without a meeting bot, while dictation and contextual text tools also run on device.
Shared workspaces: The Docker version supports multiple users with permissions and an embeddable chat widget; a developer API connects workspaces to other applications.

The free desktop app runs on Windows, macOS and Linux. Android runs small models on the phone and can connect to desktop workspaces and larger models. Docker supports deployment on your own server, and a hosted instance is also available. The project is open source under the MIT license.

Local models and embeddings can keep document processing on your device. Cloud model requests and web search use external services. Optional telemetry sends usage events, including chats and document changes, without their contents.