How Dhravya Shah Built Supermemory and Raised $3M for AI Memory

Supermemory founder Dhravya Shah took an unconventional route into AI infrastructure, turning a personal knowledge-management experiment into a startup designed to help AI applications retain useful context over time. The company announced a $3 million pre-seed round in October 2025, led by Susa Ventures alongside Browder Capital and SF1.vc, with participation from several technology executives and founders. The funding is therefore not a new 2026 round, despite renewed attention around Shah’s story this week.

Supermemory began as a consumer “second brain” project for organising information collected across different applications. Shah later saw a larger opportunity in providing the underlying memory infrastructure to developers building AI products. The platform is designed to ingest information from sources such as documents, chats and other digital content, then retrieve relevant context when an AI application needs it.

Before founding the company, Shah worked with Cloudflare’s Workers AI team and built numerous software projects while studying in the US. Reports have sometimes described him as an IIT dropout, but Shah has publicly corrected that account: he prepared for IIT entrance examinations but did not attend IIT. He later left college in the US to focus on Supermemory and moved to San Francisco. Shah also said when announcing the funding in October 2025 that he had recently turned 20.

The startup’s investor group includes industry figures connected with companies such as Cloudflare and Google DeepMind. Supermemory said the capital would support development of its memory engine for large language models and AI agents, where persistent context remains a significant infrastructure challenge.

The product strategy has continued to evolve. In September 2026, Supermemory discontinued its Company Brain product and Nova consumer platform, refunded paying users and said itwould focus more heavily on its core memory API and infrastructure layer. That shift makes the company’s central bet clearer: rather than building another standalone AI assistant, Supermemory wants to become the memory system other AI products build on.