Mem0押注每个AI智能体都需要一个可移植的记忆层

做的是什么生意

Open-source memory layer for AI agents: an API and SDK that stores, retrieves and evolves user memory across models, apps and agents, plus a managed cloud platform.

启动资金：$24M total: $3.9M previously unannounced seed plus $20M Series A led by Basis Set Ventures, with Peak XV Partners, Kindred Ventures, GitHub Fund and Y Combinator (announced Oct 28, 2025)

起因

Taranjeet Singh, an ex-Paytm and Khatabook engineer who built Embedchain, teamed up with Deshraj Yadav, who had led the AI platform at Tesla Autopilot, and launched Mem0 in January 2024. While experimenting with Embedchain they built a meditation app that went viral in India, and the recurring user complaint — 'the app doesn't remember me' — became the product.

经过

Mem0 grew through open source and Y Combinator: the framework integrated with LangChain, LlamaIndex, CrewAI, Flowise and Langflow, and AWS selected Mem0 as the memory provider for its Agent SDK. By the funding announcement in October 2025 the open-source project had surpassed 41,000 GitHub stars and 14 million Python downloads, API volume jumped from 35 million calls in Q1 2025 to 186 million in Q3 2025, and more than 80,000 developers had signed up for the cloud service. The $24M round — $3.9M seed plus $20M Series A led by Basis Set Ventures — brought in Peak XV Partners, GitHub Fund and angels like Dharmesh Shah, Scott Belsky and Olivier Pomel.

结果

Scaling: with $24M, Mem0 plans to expand its engineering team, harden memory for enterprise use cases and build partnerships with AI platforms; the GitHub repo remained actively developed through 2026, shipping a new memory algorithm in April 2026.

可借鉴之处

在用户抱怨处转向：来自爆红副业项目的反馈——‘它记不住我’——暴露了比应用本身更大的平台缺口，为这一痛点构建的记忆层成就了2400万美元的业务。