AnythingLLM
Self-hosted AI assistant for chatting with documents. Desktop app needs no Docker or server setup. Multi-user Docker version adds workspace isolation and role-based access. 30+ LLM providers
Quick Start
docker run -d -p 3001:3001 mintplexlabs/anythingllm Overview
AnythingLLM is a self-hosted AI assistant built around chatting with documents. Drop in a PDF, a Word file, a codebase, or a set of URLs and it parses, chunks, and indexes the content so you can ask questions against it. The workspace model keeps different document sets separate — useful when you want one workspace for contracts, another for technical documentation, and another for a specific project.
The standout feature for non-technical users is the desktop application. Install it on macOS, Windows, or Linux and you get the full experience without Docker, without a server, without any configuration beyond choosing an LLM provider. Point it at a local Ollama instance or connect an OpenAI or Anthropic API key, and it works.
For team deployments, the Docker version adds multi-user support with role-based access, workspace isolation, and an admin panel — the governance layer that Open WebUI lacks. Over 30 LLM providers are supported including OpenAI, Anthropic, Ollama, LM Studio, Azure, AWS Bedrock, and Groq.
The honest limitations: AnythingLLM is reactive. It waits to be asked. It will not surface items from your documents unprompted, send notifications, or take autonomous actions outside the chat window. If you need agent workflows that connect to external services, that is a different tool.
For individuals or small teams who want private document Q&A without a complex setup, this is the most approachable option in the category.
AnythingLLM: Pros & Cons
| Pros (The Wins) | Cons (The Friction) |
|---|---|
| Desktop app: Full experience on macOS, Windows, Linux — no Docker needed. | Reactive only: Waits to be asked; no proactive alerts or autonomous actions. |
| Document RAG: PDF, Word, CSV, URLs indexed and queryable per workspace. | Cloud pricing: Hosted plan starts high; self-host or desktop to avoid it. |
| 30+ LLM providers: Ollama, OpenAI, Anthropic, Groq, AWS Bedrock, and more. | Agent limits: Cannot send emails, book meetings, or connect to external services. |
| Multi-user governance: Workspace isolation, RBAC, admin panel in Docker version. | RAG quality varies: Large or poorly formatted PDFs produce inconsistent results. |
Use Cases
Specific ways to use AnythingLLM for your workflow.
Deployment Strategy
Recommended ways to host AnythingLLM in your own environment.