What it does
Mem0 adds a memory layer to AI assistants and agents so they can remember user preferences and adapt over time. It keeps user, session and agent state, and the repository names customer support bots, AI assistants, healthcare and gaming as example uses. There are three ways to run it. The library installs with pip or npm, a self-hosted server starts with docker compose and includes a dashboard and API key authentication, and a hosted Cloud Platform is reached by signing up and using SDKs or API keys. A command-line tool lets you add and search memories from a terminal, and the page says AI agents can sign themselves up for an API key from the CLI. Mem0 needs a language model and an embedding model to work. The page names two OpenAI models as defaults and says many others are supported. Retrieval combines semantic search, keyword matching and entity matching, and memory extraction runs in a single pass. The project notes that its published benchmark scores come from the managed platform, which has optimizations not in the open-source SDK, and that the evaluation framework is open-sourced. Integration examples cover a ChatGPT-style demo, a browser extension, LangGraph and CrewAI, and the project offers skills that teach coding assistants such as Claude Code, Codex and Cursor how to use it. The code is licensed under Apache 2.0.
Capabilities
- Memory for user, session and agent state
- Python and npm libraries plus a command-line tool
- Self-hosted server via docker compose, or a hosted Cloud Platform
- Retrieval that combines semantic, keyword and entity matching
- Examples for LangGraph, CrewAI and a browser extension
- Skills for coding assistants such as Claude Code, Codex and Cursor
When to use it
- You are building a chatbot or agent that should recall users across sessions
- You want memory infrastructure you can host yourself under an open-source licence
- You prefer a hosted memory service and no servers to run
- You build customer support assistants that need past tickets and history
