Mem0 is an open-source memory layer for AI agents that solves the problem of stateless LLMs by storing and retrieving user interactions so an assistant can improve and personalize its responses over repeated sessions. Taranjeet Singh and Deshraj Yadav launched the company in January 2024 through Y Combinator, and it has since raised $24 million, including a $20 million Series A led by Basis Set Ventures with participation from Peak XV Partners and the GitHub Fund.
Rather than stuffing an entire conversation history into a model's context window, Mem0 selectively extracts and stores what's worth remembering, then retrieves the relevant pieces on each new call, which the company positions as both cheaper and more accurate than naive context-window approaches. The open-source core is available on GitHub, and the hosted cloud API, used by more than 80,000 developers, now serves as the default memory provider for AWS's Agent SDK.
Mem0 targets developers building AI agents and chat products who need persistent memory without building storage and retrieval infrastructure themselves. A free Hobby tier covers 10,000 memory adds and 1,000 retrieval calls a month; paid plans start at $19 a month for the Starter tier and scale to $249 a month for Pro, with usage-based enterprise pricing above that.
