Mem0
An AI agent memory layer with Apache 2.0 self-hosted code and a managed service. It recalls user and agent context across sessions.
Mem0 is a memory layer for developers building AI agents and assistants that need to recall earlier interactions. It retains user, session, and agent context across conversations, so an assistant can remember preferences or a support bot can refer to past tickets. The self-hosted code is open source under Apache 2.0; Mem0 also has a managed service.
Mem0 turns interactions into compact memories and retrieves relevant details when an agent needs them. It can retain facts an agent records after completing an action, link references to the same entity, and distinguish past events from current information. Retrieval combines semantic similarity, keyword matching, and entity links. Compact memories also reduce the amount of chat history an app needs to send to a model.
The self-hosted server can run on Kubernetes, in a private cloud, or in an air-gapped environment. It still needs a language model and an embedding model. The documented defaults use OpenAI models for generation and embeddings, so that setup sends model requests to OpenAI. Fully local operation requires local model and embedding providers. Python and Node.js SDKs are available, with integrations for LangGraph and CrewAI. A Chrome extension stores memories across ChatGPT, Claude, and Perplexity.
The managed platform includes proprietary retrieval optimizations that aren't in the open source code. Mem0's enterprise offering also logs memory reads and writes and supports bring your own key controls.
