At 19, Dhravya Shah has built an AI startup around a problem that could become very important as AI tools become more useful: memory.
His company, Supermemory, has raised $3 million in funding. The startup wants to help AI systems remember useful information from past conversations, files, emails and other digital sources. Its technology can give AI tools a form of long-term memory, so they do not have to treat every new conversation as if it starts from zero.
The news has drawn fresh attention on September 29, 2026, after reports highlighted Shah’s unusual path from software projects to startup founder. His story stands out because he chose to leave college and focus on AI products at a very young age.
Supermemory itself is not a brand-new company. Shah founded it in April 2025, and the company announced its $3 million round in October 2025. Today’s reports have brought that story back into focus because of Shah’s age, his background and the growing need for better memory in AI systems.
What Supermemory does
The basic idea behind Supermemory is simple.
AI models can be very smart, but they do not always remember a user from one session to the next. A person may tell an AI tool something important today, then have to explain the same detail again later.
Supermemory aims to solve that problem.
The company describes its product as a memory engine for large language models and AI agents. It can take information from different sources and turn it into useful context for an AI system. That information can include emails, chats, files, PDFs and documents.
The platform can also connect with services such as Gmail, Google Drive, Notion and GitHub. This lets it collect and update information from places where a person or company already stores data.
The aim is not simply to save information. The system must also find the right detail when an AI application needs it.
That is where the technology becomes more complex.
Why memory matters for AI
Most people have seen how AI can answer questions, write text, create code or work with documents. But useful AI needs more than strong answers.
It also needs context.
Imagine a user who has worked with an AI assistant for six months. The assistant may know a lot about the user’s work during one conversation, but it can lose that context when a new session starts.
A memory system can help solve this gap.
It can store useful details and bring them back when they matter. This could allow an AI assistant to understand a user’s preferences, past decisions, work habits and important information without asking the same questions again.
For companies, the value can be even greater. An AI agent may need to remember information about customers, projects, software code or company documents.
Supermemory wants to provide that memory as infrastructure that developers can add to their own AI products.
From a simple idea to a startup
Supermemory did not start as a large business idea.
Shah first created it as a consumer product. It began as an open-source “second brain” app that could help people organise knowledge from different apps. The project gained interest from users and developers.
Some users were ready to pay for it. Others wanted help with the open-source version.
That gave Shah a clue that there could be a larger business opportunity.
Instead of only building a product for individual users, he moved toward infrastructure for developers. The goal became much broader: give AI applications a reliable memory layer that can work across different models and tools.
That change helped turn Supermemory from a side project into a startup.
Shah started young
Shah’s interest in software began before Supermemory.
According to reports, he worked on software projects while he prepared for engineering entrance exams. One of his early projects earned enough money to help him move to the United States. He later received an O-1 visa, a US visa for people who show extraordinary ability or achievement in areas such as science, education, business, arts or athletics.
He also followed a very unusual approach to product development.
Instead of waiting for one perfect idea, Shah set himself a challenge to build one project every week for 40 weeks. That gave him a chance to test many different ideas and learn what people found useful.
One of those projects eventually led to Supermemory.
This approach also helped him learn what it takes to create software that people actually want to use.
His time at Cloudflare
Before Supermemory became his main focus, Shah worked at Cloudflare.
He joined the company as an engineering intern with its Workers AI team in Arizona. He later moved into a developer relations role.
His work gave him exposure to AI infrastructure and the tools that developers use to create AI applications.
During his internship, he worked on function calling for an AI platform. He also filed a patent related to that technology, according to details cited in reports about his career.
He also created an open-source project for AI agents. Another project involved a Go script that helped remove millions of phishing and scam domains from Cloudflare, according to information from his profile cited by Jagran Josh.
His developer relations work also gave him direct contact with software developers. He visited colleges, took part in hackathons and conferences, and showed products to developers.
That experience likely gave him a useful view of the problems faced by people who build AI tools.
A different path from college
Shah’s education path has received a lot of attention.
He began a bachelor’s degree in computer science in 2023 but later left the programme. Reports have described him as an IIT dropout, but his own profile disputes that description and says he did not attend IIT Bombay.
That distinction matters because several reports have used different descriptions of his education.
What is clear is that Shah chose to leave the traditional college path and focus on software and AI.
His decision does not mean that college is unnecessary for everyone. It simply forms part of his own story. His experience shows how some young technology founders choose a different route when they believe they have found a problem worth solving.
In Shah’s case, that path led to Supermemory.
The company raised $3 million
Supermemory’s funding round was announced in October 2025.
The company said it raised $3 million in its first funding round. The round was led by Susa Ventures, Browder Capital and SF1.vc. Several well-known technology figures also took part as angel investors.
The list included people linked to Google, Cloudflare and other technology companies. Reports have highlighted names such as Google AI chief Jeff Dean, Cloudflare CTO Dane Knecht and DeepMind product manager Logan Kilpatrick.
There is a small difference in how the funding has been reported. Supermemory itself describes the round as $3 million, while some earlier reports put the figure at about $2.6 million. For this article, the company’s own announcement and the current reports use the $3 million figure.
The funding gives the company resources to improve its technology and expand its work with developers and businesses.
Investors see a larger problem
The interest from experienced technology investors is also important.
AI has made major progress in recent years. Models can write code, answer questions, analyse information and work with many forms of data.
But memory remains a difficult problem.
A model may know a lot from its training data, yet that does not mean it knows a particular user’s history. It may also struggle to keep track of information across long periods.
Supermemory’s goal is to provide a separate memory layer that can handle this problem.
The company says its system includes its own vector database, content parser and extractor. These parts help it process information, store it and later find relevant details.
In simple terms, Supermemory wants to act like a memory system that sits behind an AI application.
From consumer app to AI infrastructure
One of the most important parts of Supermemory’s story is its shift from a consumer app to developer infrastructure.
At first, the product helped people collect and organise information. But Shah saw that developers also needed a way to add memory to their own AI products.
That created a much bigger market opportunity.
An individual user may need one memory system. Thousands of AI applications may need one too.
Supermemory therefore opened its technology for developers. Its memory engine can help AI applications store context and retrieve it when needed.
This approach fits a larger change in the AI industry. Developers increasingly use separate tools for models, databases, agents, search and memory.
Supermemory wants to become one of those core tools.
The company has continued to evolve
Supermemory’s work did not stop after its first funding round.
The company’s own site shows several product changes in 2026. In September, Supermemory said it had discontinued its company brain and Nova products. It said customers who had paid for those products received refunds, while its MCP and plugins continued to operate. The company also said it would focus fully on its memory engine.
This is an important detail because startups often change their products as they learn more about the market.
Supermemory’s latest direction is clear: focus on memory and context infrastructure for AI models and agents.
The company now describes its technology as memory and continual learning infrastructure that can work with different models and AI systems.
Why this could matter for AI agents
AI agents are becoming more capable of taking several steps to complete a task.
But an agent without memory can have limits.
Suppose an AI agent helps a person manage software projects. It may need to remember past decisions, code changes, customer requests and personal preferences.
If it forgets those details, the user has to provide the same information again.
A memory layer can reduce that problem.
Supermemory aims to give developers a way to add this ability without building the entire system themselves. Its technology can store information, organise it and bring relevant context back when an AI application needs it.
This could become more important as AI agents take on longer and more complex tasks.
A bigger race for AI infrastructure
Supermemory is part of a wider AI infrastructure market.
Large technology companies are spending huge sums on AI models and data centers. At the same time, smaller startups are building specialised tools that support those models.
Memory is one such area.
The need is easy to understand. A powerful AI model can answer a question well, but an application becomes much more useful when it also knows what happened before.
That is the problem Supermemory wants to solve.
Its challenge now is to make that memory reliable, fast and useful at a large scale. It also has to work across different AI models and applications.
What comes next for Supermemory
The $3 million round gives Shah and his team a base for the next stage of the company.
Supermemory has already moved from a small open-source project to an AI infrastructure company. Its technology now targets developers and businesses that want memory for their AI applications.
The company has also built relationships with enterprise customers and open-source projects. In its 2025 funding announcement, Supermemory said some customers were sending billions of tokens through the system each week. It also named companies such as Cluely and Composio among its enterprise customers, along with Scira AI as an open-source user.
The next test will be whether Supermemory can turn this early interest into a large and lasting business.
For Shah, the journey is already unusual. At 19, he has built a company around one of AI’s less visible but important problems. His focus is not on creating another chatbot. Instead, he wants to give AI systems something closer to long-term memory.
As AI tools become more personal and more capable, that problem could become even more important. Supermemory’s $3 million raise gives the young founder more resources to pursue that idea, while its shift toward memory infrastructure places the company in a part of the AI market that may become a key layer for future AI applications.
Also Read – AI Leads the New Wave of Startup Technology