In Silicon Valley, headlines are dominated by the same few names—OpenAI, Anthropic and Google—raising billions of dollars and building products that unsettle startups trying to build solutions on top of them. For a young founder, finding a moat in the AI space feels extremely difficult. For a solo founder, it’s near unthinkable.
Yet, Dhravya Shah from Mumbai was just shy of 20 when he raised $3 million for his solo startup, Supermemory. At an age when most students are still figuring out internships or college majors, Shah has built a company valued at $40 million, attracting investments from Google AI chief Jeff Dean, Cloudflare CTO Dane Knecht and DeepMind’s Logan Kilpatrick, among other top Silicon Valley names.
He’s done it without a co-founder. Just a laptop, a vision, and an uncanny ability to code.
Supermemory is designed as a universal memory layer for AI applications. The product takes unstructured data—files, chats, emails, project updates, PDFs—and converts them into a personalised knowledge graph for users across different AI tools like ChatGPT, Gemini and Claude. These apps can then recall this memory to connect relevant information across time and platforms, whether it’s an old note in a journaling app or visual assets for a video editor.
In short, it offers AI systems memory beyond a single session or platform, a gap that even the most advanced models from Google and OpenAI haven’t been able to fill.
“I still would like a co-founder. I’m neither great at sales nor engineering. I do everything myself, which is fine,” Shah told AIM. He draws inspiration from the ‘solo together’ concept coined by Julian Weisser at ODF. In this setup, solo founders live together and work side by side, helping each other with strategy, decisions and direction while building their own companies.
“There are a lot of things you need a co-founder for, but usually things that are not directly related to tech—co-founders of convenience,” he said. “It’s a really cool concept.”
The Journey
Shah’s journey has been anything but conventional. He started programming at 16, building games before realising coding was more than a hobby. “Three months later, I got into Python and found the conviction to learn coding seriously,” he said.
By 17, he had already built a Twitter bot, Tweets.beauty, which generated images from tweets and attracted more than five million users, until soaring server costs forced him to shut it down. He sold the underlying technology to HypeFury, using the proceeds to fund his move to Arizona State University, abandoning earlier plans to pursue IIT.
However, he is quick to distance himself from the media narratives around IIT. “I have no heritage, no flashy background. I went to a bad school in a suburb—Nirmala Memorial College in Mumbai. I didn’t go to IIT. People assume it adds value, but it’s just noise.”
By the time he was 18, Shah was building Radish, an open-source alternative to Redis. When Redis changed its license, restricting commercial use, a devastated Shah decided to rebuild it from scratch. “I started with a normal dictionary, then made commands like ‘get’ and ‘set’, then figured out the Redis protocol. In just a few hours, I had a working Redis server,” he explained.
Radish isn’t just a clone; it’s designed to be open and customisable. Shah works in sprints across multiple projects, never focusing on just one thing. “I can’t concentrate on just one thing,” he admitted. Radish was one of over 60 products he built, ranging from social media platforms to image-sharing websites.
Supermemory itself began as Any Context on GitHub, a tool to chat with Twitter bookmarks. It has since evolved into a multimodal platform integrated with Google Drive, OneDrive and Notion, and is accessible via a Chrome extension. Its chatbot and note-taking features let users add files, links and text-based notes that become “memories” for AI systems to use.
Shah’s early customers include startups like AI video editor Montra, AI search engine Scira AI, real estate tech firm Rets, a16z-backed desktop assistant Cluely, and a robotics company using it to retain visual data. Even so, he says his long-term goal has always been to create a consumer app.
Everyone Wants a Slice of Supermemory
The investment story, Shah insists, is almost accidental. “I did not approach any of these investors.” The Google and DeepMind executives came in only after Shah had turned down an offer from DeepMind and rejoined Cloudflare.
A Twitter connection spread the word that he was raising funds, and suddenly, he had more investors than he could have imagined. Even Y Combinator came knocking, but he declined as he already had investors and a validated product. “YC is amazing for people just starting, pre-product. I’d been building Supermemory for two years. I had an audience, but no revenue because of my F1 visa.”
Today, on an O-1 extraordinary ability visa, Shah can build the business without restrictions.
The core challenge Supermemory is addressing is as fundamental as it is unsolved: interoperability and memory in AI. Today, developers juggle multiple AI models, each with its own memory—or none at all. Consumers switch between chatbots with siloed memories.
Shah’s solution sits above it all, as a universal memory layer that integrates across models and platforms, providing context where none existed before. “It has to be in the routing layer, not inside model providers. That’s the interesting part,” he said.
Users can store memory in Supermemory’s cloud or their own cloud, with the Chrome extension or API allowing the memory to follow them wherever they work, not just inside a standalone app.
Shah describes the product as part developer tool, part consumer application. While the immediate focus is on the developer and enterprise side—millions of startups need memory layers—he envisions a consumer base that could enable features like ‘login with Supermemory’.
Making enterprise integrations worthwhile requires enough users to dominate the market. “Indian VCs come after you’re validated by American VCs. They will not take the first bet. I have three or four Indian VCs who invested. They’re thoughtful, helpful and give introductions. But they won’t make big, bold bets.”
Unique Pressures
Being solo comes with its fair share of concern. Shah pointed out that everyone in his team so far, including engineers and designers, has been under 20. “The successful people would cost $300,000 a year; I would run out of money,” he said.
He relies heavily on Twitter to build the brand and attract talent. He works more than other founders, tries to avoid distractions and insists on doing his own thinking rather than outsourcing it to AI.
“I learned to code before AI became mainstream. It’s like a drug—always there. Learning to code yourself is irreplaceable. Otherwise, people write bad code relying on AI,” he said.
Shah’s approach to AI memory and finding the moat in the market is deliberately independent of giants like Google or OpenAI. “If each model provider adds their own memory, developers lose interoperability. Consumers use different chatbots, each with their own memory. These providers aren’t incentivised to integrate. Supermemory becomes that independent layer,” he explained.
The product’s appeal lies in both speed and flexibility. Competing startups exist—Letta, Mem0, Memories.ai—but Supermemory prioritises low latency and universal compatibility.
Shah’s current focus is on expansion and scaling. ‘Login with Supermemory’ is the next big milestone. He wants to dominate both the consumer and enterprise spaces, ensuring enough adoption to make the integrations compelling.
Beyond that, his mind drifts to big ideas—AI infrastructure, space, even social products tackling loneliness. But for now, what keeps him awake is operational: keeping customers happy as Supermemory scales rapidly.
Even at 20, Shah seems clear-eyed about the fact that building a successful company is about more than code. Investor expectations for solo founders are clear: hire thoughtfully, build a brand people want to work for and deliver. Supermemory’s story is proof that age, co-founders and traditional pedigree are not prerequisites for building something world-changing.
As for the future, Shah revealed Supermemory is less about hype and more about impact. He wants to take 10 big enterprises, keep them extremely happy before considering an eventual IPO. Media noise, inflated headlines about IIT or Google-level competition hold little value to him.
“That adds nothing to my life. Except followers,” he said with a laugh.
The post The 20-Year-Old Indian Prodigy Who Gave AI a Supermemory appeared first on Analytics India Magazine.
