Taranjeet Singh, a serial entrepreneur with a history of launching and scaling startups, has unveiled Mem0, a new AI company focused on solving one of the most pressing limitations of large language models: the inability to remember past interactions. Mem0 aims to create a persistent, portable memory layer for AI applications—what Singh calls a “memory passport”—allowing AI agents to retain and build upon user history across different apps and platforms, much like how login credentials or email accounts follow users today. Launched in January 2024, Mem0 has raised $24 million in a combination of previously unannounced seed funding and a Series A round led by AI-focused early-stage fund Basis Set Ventures. Additional investors include Y Combinator, Peak XV Partners, the GitHub Fund, and existing backers Kindred Ventures. The company has also attracted notable angel investors such as Dharmesh Shah (HubSpot), Scott Belsky (ex-CPO at Adobe), Olivier Pomel (Datadog), Thomas Dohmke (ex-CEO of GitHub), Paul Copplestone (Supabase), James Hawkins (PostHog), Lukas Biewald (Weights & Biases), Brian Balfour (Reforge), Philip Rathle (Neo4j), and Jennifer Taylor (former President of Plaid). Despite its small team of just four, Mem0 has seen rapid traction. Its open-source API, which enables developers to build memory into AI applications, has surpassed 41,000 GitHub stars and recorded over 13 million Python package downloads. In Q1 2025, the platform processed 35 million API calls; by Q3, that number surged to 186 million—representing a monthly growth rate of around 30%. More than 80,000 developers have signed up for its cloud service, which now handles more memory operations than any other provider and is the exclusive memory layer for AWS’s new Agent SDK. Singh’s journey to Mem0 began in Bangalore, where he worked as a software engineer at Paytm and later became Khatabook’s first growth engineer. After leaving in late 2022, just before the ChatGPT explosion, he built one of the first GPT app stores, which grew to over a million users. This experience led him to create Embedchain, an open-source tool for indexing and retrieving unstructured data, which gained over 8,000 GitHub stars. While working on Embedchain, Singh and his co-founder and CTO Deshraj Yadav—formerly head of the AI Platform at Tesla Autopilot—launched a meditation app inspired by Indian spiritual leader Sadhguru. The app went viral in India, but users repeatedly asked the same thing: “I’m on a journey, but the app doesn’t remember me.” That feedback sparked the pivot to Mem0. While other AI labs like OpenAI are developing internal memory systems, Singh argues they have little incentive to make them open or interoperable. “Memory is becoming a key moat as LLMs become commoditized,” he said. Mem0’s goal is to be a neutral, open infrastructure layer—what Singh describes as “Plaid for memory”—so developers can build truly personalized AI experiences without being locked into a single platform. The framework is model-agnostic, supporting OpenAI, Anthropic, and any open-source LLM, and integrates with popular tools like LangChain and LlamaIndex. It enables applications such as therapy bots that recall past sessions, productivity agents that learn user habits, and AI companions that evolve over time. “Memory is foundational to the future of AI,” said Lan Xuezhao, founder and partner at Basis Set Ventures. “We’re backing Mem0 because it’s tackling one of the hardest and most important infrastructure challenges in the AI stack.” Mem0 is entering a growing market, with other startups like Supermemory, Letta, and Memories.ai also building in the memory space. But with strong early adoption, top-tier backing, and a clear vision for open, portable memory, Mem0 is positioning itself as a critical piece of the next generation of AI infrastructure.