AI models’ contextual windows, which indicate a model’s ability to “remember” information, have increased over time. However, scientists suggested recent ways To increase the long-term memory of AI models, as they often cannot have context across several sessions.
Nineteen-year-old founder Dhravya Shah is trying to resolve problems in this area by building a memory solution called Supermemoriesfor AI applications.
Originally from Mumbai, India, Shah began building bots and consumer apps a few years ago. He even sold his bot, which formatted tweets into handsome screenshots, to social media tool Hypefury.
The founder, who was preparing for the entrance exam to get into IIT (Indian Institute of Technology), made good money from this sale and decided to maneuver to the US to attend Arizona State University as an alternative.
After moving for 40 weeks, he threw himself into building something recent for 40 weeks. During one of those weeks, he built Supermemory (which was initially called FreeContext) and put it on Github. During this time, the tool allowed you to talk with your Twitter bookmarks.
The current version of the tool extracts “Memories” or insights from unstructured data and helps applications higher understand context.
Shah secured an internship at Cloudflare in 2024, where he worked on artificial intelligence and infrastructure. He later worked as the company’s chief developer relations officer. At that point, advisors including Cloudflare CTO Dane Knecht asked him to show Supermemory into a product.
This yr he decided to build Supermemory full time.
Supermemory, currently described as a universal memory API for AI applications, builds a knowledge graph based on the data it processes and personalizes the context for users. For example, it could handle queries in monthly writing entries or journaling applications, or searches for email applications. Since the solution allows for multimodal inputs, it could also allow the video editor to drag appropriate assets from the library for a specific prompt.
The startup can devour any form of data, the company says, including files, documents, chats, projects, emails, PDFs and application data streams. The Chatbot and Notetaker feature allows users so as to add memories to text, add files or links, and hook up with apps like Google Drive, OneDrive or Concept. There is also a Chrome extension that means that you can easily add notes from a website.
“Our core strength is extracting insights from any unstructured data and giving applications more user context. As we work on multimodal data, our solution is suitable for all kinds of AI applications, from e-mail clients to video editors,” Shah said.
Supermemory has secured $2.6 million in seed funding led by SUSA Ventures, Browder Capital and SF1.VC. The round also includes individual investors similar to Cloudflare’s Knecht, Google AI chief Jeff Dean, DeepMind product manager Logan Kilpatrick, Sentry founder David Cramer, and executives from Openai, Meta and Google.
Shah said that at one point Y Combinator also approached him to hitch one of its lots, but it already had investors on board, so the timing didn’t work out.
Joshua Browder, founder and CEO of robot lawyer startup Donotpay, who runs Browder Capital as a solo GP, was impressed by Shah’s persistence.
“I connected with Dhravya through X, and I was struck by how quickly he moves and builds things, and that made me invest in him,” he said.
The company has a variety of existing clients, including A16Z-powered desktop assistant Cluely, AI video editor Montra, AI Search ScreamingComposio’s Multi-MCP Rube tooland real estate startup Rare. Additionally, it is working with a robotics company to preserve visual memories captured by the robot.
While this is skewed towards consumers, the app is more like a playground for developers to grasp more about the tool and potentially use it in their workflows or their very own applications.
Supermemory has significant competitors in the memory space. Startups like Felicis Ventures backed by Letta and MEM0 (where Shah worked for a short time) are building a storage layer for agents. Susa Ventures, Supermemory’s own backer, has invested in Memories.ai with Samsung, which might leverage hundreds of hours of footage to realize insights. Shah says that while these startups may serve different industries and use cases, Supermemory will stand out because it offers lower latency.
“More and more AI companies will need a memory layer. SuperMemory delivers high performance while enabling rapid persistence of the appropriate context,” Browder said.