Mem0

Mem0 is a universal memory layer for LLM apps and AI agents. It extracts, stores, and recalls the facts and preferences that matter—so your agent can personalize behavior across sessions while using fewer tokens. Developers can use the managed cloud or the open-source SDKs (Python/Node) to add working, episodic, semantic, and factual memory, with sub-50 ms lookups, decay/filters to prevent memory bloat, and optional graph memory to connect entities over time. Mem0 integrates with popular agent frameworks (LangChain, CrewAI, LangGraph, AutoGen, Vercel AI SDK, etc.) and supports MCP via an official example and the OpenMemory MCP app. A hosted dashboard provides observability and analytics. Security is covered via a dedicated trust portal, with SOC 2 Type I and HIPAA compliance listed and SOC 2 Type II in observation.

Key use cases: persistent user profiles, cross-session personalization, memory for support/chat, voice agents, and hybrid RAG + memory setups.