The windows of the context have increased from artificial intelligence models, which indicate the model’s ability to “remember” information, over time. However, the researchers have Proposal New ways To increase the long -term memory of artificial intelligence models, because it often cannot keep the context on several sessions.
Its nineteen -year -old founder Dhravya Shah is trying to solve problems in this field by building a memory solution, called SuperMoryFor artificial intelligence applications.
Originally from Mumbai in India, Shah began building robots and applications for consumer a few years ago. He even sold his robe that has been coordinated in well -looking screenshots to HIPEFURY.
The founder, who was preparing for the admission exam to enter IIT (Indian Institute of Technology), got good money from this sale and decided to move to the United States to attend the Arizona State University instead.
After moving, he challenged himself to build something new every week for 40 weeks. During one of those weeks, SupermeMory (which was initially called any context) built and placed it on Gaytap. At that time, the tool allowed you to chat with your Twitter reference.
The current version of the tool is extracted from the tool “memories” or visions of the non -structured data and helps applications to better understand the context.

Shah got internal training in Cloudflare in 2024, where he worked on artificial intelligence and infrastructure. He later worked as a developer relations leader in the company. During this time, consultants, including Cloudflare Cto Dane Knecht, asked him to convert the bone into a product.
This year, he decided to build a full -time memory.
SuperMory, now described as a global memory application programming interface for artificial intelligence applications, builds a graphic fee for knowledge based on the data it processes, and customize the context of users. For example, you can support query via a month’s age to get an application for writing, a diary app, or search for an email application. Since the solution allows multimedia input, the video editor may also be allowed to bring in relevant assets from a library for a specific demand.
The company says that the startup can accommodate any type of data, including files, documents, conversations, projects, email messages, application data flows and application data flows. Chatbot and Notetaker feature allow users to add memories in the text and add files or links, and contact applications such as Google Drive, OneDrive or idea. There is also an extension of the chrome that allows you to easily add notes from the website.
Shah said: “Our basic strength is to extract visions of any kind of non -structured data and give applications more context about users. Since we are working through multimedia data, our solution is suitable for all types of artificial intelligence applications ranging from email customers to video editors.”
SupermeMory received the 2.6 million dollar seed financing led by Susa Ventures, Browder Capital and SF1.VC. The tour also includes individual investors such as Cloudflare’s Knecht, Google Ai Jeff Dean, Director of DeepMind Logan Kilpatrick Products, Sentry David Cramer Products, Openai, Meta and Google.
Shah said at some point, as Y Combinator approached him to join one of his batches, but he had already had investors on board, so the timing did not succeed.

Joshua Broder, founder and executive of the company “Robot Lawyer”, was influenced by Browder Capital as a Solo GP, with Shah’s perseverance.
He said: “I contacted Dhravya on X, and what surprised me is how quickly it moves and builds things, which led me to invest in it.”
The company has many current customers, including A16z Cluely desktop assistantArtificial Intelligence Video Editor MontraSearch artificial intelligence SciraComposio Multi-MCP Rube toolAnd start real estate Ritz. In addition, it works with a robotics company to keep the visual memories captured by a robot.
Although this is inclined to consumers, the application is similar to the stadium for developers to understand more about the tool and perhaps use it in the workflow or their own applications.
SuperMory has large competitors in the memory space. Startups Letta subsidized. And Mem0 (where Shah worked for a short time) build a memory layer for agents. Private SupermeMory supporter, Susa Ventures, invested in Memories.ai along with SamsungWhich can benefit from thousands of hours from the shots for visions. Shah says that although these startups may serve different industries and cases of use, SupermeMory will stand out because they provide less transmission time.
“More and more artificial intelligence companies will need a memory layer. SuperMory solution provides a high performance while allowing you to quickly relevant.”
https://techcrunch.com/wp-content/uploads/2025/10/Dhravya-Shah-Supermemory.jpeg?resize=1200,800
Source link