The ambient windows of AI models, which indicate the ability of a model to “remember” the information, have increased over time. However, researchers have proposed new ways To increase the long -term memory of AI models, as they often cannot keep the frame in various sessions.
Nineteen -year -old founder Dhravya Shah is trying to solve problems in this area by creating a memory solution called SuperstarFor AI applications.
Initially from Mumbai in India, Shah started making bots and applications facing consumers a few years ago. He even sold his bot that formed the tweets in beautiful screenshots on the hypefury tools of social media.
The founder, who was preparing for entry to enter the IIT (Indian Institute of Technology), earned good money from this sale and decided to move to the US to attend the State University of Arizona.
After the relocation, he questioned himself to build something new every week for 40 weeks. During one of these weeks, he built the supermemory (which was originally named any frame) and put it on Github. At that time, the tool allowed you to chat with Twitter bookmarks.
The current version of the tool extracts “memories” or ideas from unstructured data and helps applications better understand the box.
Shah secured an internship at Cloudflare in 2024, where he worked on AI and the infrastructure. Later he worked as a developer relations in the company. During this time, consultants, including Cloudflare Cto Dane Knecht, asked him to convert Supermemory into a product.
This year, he decided to build the supermemory full -time.
Supermemory, now described as API Universal Memory for AI applications, creates a knowledge graph based on data that processes and personalizes the box for users. For example, it can support the search for all month entries for a writing or journalism application or look for an email app. As the solution allows multimodal entrances, it could also allow a video processor to transfer relatively assets from a library for a particular prompt.
Startup can consume any type of data, the company says, including files, documents, conversations, projects, e -mail, PDF and applications. Chatbot and Notetaker allows users to add text memories, add files or links, and connect to applications such as Google Drive, OneDrive or Invion. There is also a Chrome extension that allows you to easily add notes from a website.
“Our basic power is to extract knowledge from all kinds of unstructured data and give applications more frames for users. As we work in multimodal data, our solution is suitable for all types of AI applications ranging from email customers to video publishers,” Shah said.
Supermemory has secured $ 2.6 million seed funding led by Susa Ventures, Browder Capital and SF1.Vc. The round also includes individual investors such as Cloudflare’s Knecht, head of Google AI Jeff Dean, Product Manager Deepmind Logan Kilpatrick, Founder of Sentry David Cramer and executives from Openai, Meta and Google.
Shah said at a point Y Combinator also approached him to join one of his lots, but he already had investors on the ship, so the timetable did not work.
Joshua Browder, founder and chief executive of the Donotpay start -up, managing the Browder Capital as a GP solo, was impressed by Shah’s durability.
“I joined Dhravya over X, and what hit me was how fast it moves and builds things and it prompted me to invest in it,” he said.
The company has many existing customers, including the assistant desktop supported by A16Z, CLUELY, AI VIDEO EDITOR MONTRA, AI Search SkiraComposio’s Multi-MCP Rube tooland starting real estate Queer. In addition, he works with a robotics company to maintain visual memories captured by a robot.
While this is a inclination to consumers, the application feels more like a playground for developers to understand more about the tool and possibly use it in their work flows or in their own applications.
Supermemory has important competitors in the memory area. The newly established companies such as Felicis Ventures-Backed Letta and MEM0 (where Shah worked for a while) make a mattress for agents. Supermemory supporter Susa Ventures himself has invested in memories. Together with Samsung, who can use thousands of hours of video to get information. Shah says that while these newly established businesses may serve different industries and use cases, the supermemory will stand out because it offers a lower latent situation.
“More and more AI companies will need a mattress. The solution of the supermemory provides high performance, while allowing you to quickly surface the frame,” Browder said.
