Most students spend their college years preparing for jobs or graduate school. Dhravya Shah passed his construction software. Before turning 20, the Indian entrepreneur had already launched more than 15 open-source projects, sold two companies, and eventually raised $3 million as a lone founder to build an AI startup that now develops memory infrastructure for intelligent agents.In a wide-ranging conversation on the Solo Founders podcast, Shah reflected on his unconventional path — from abandoning the dream of studying at IIT to dropping out of college despite campus life. Along the way, he shared why he believes that the future of artificial intelligence will depend less on smarter models and more on something that many developers have always neglected: memory and context.
The beginning was not planned – years of construction brought it
Shah says he never set out to build a startup. Instead, he spent years creating projects simply because he found them interesting. None of their first products were hidden behind a paywall, and every one of them was open source.One project, AnyContext, a tool designed to help users organize their personal context, unexpectedly gained traction. Instead of treating it as a finished product, Shah continued to listen to users and repeatedly changed direction. Over time, what began as a consumer “second brain” evolved into Supermemory, an infrastructure platform that helps developers build AI applications capable of remembering information over time.Even during fundraising, when one of his pitches generated millions of online impressions, Shah resisted the temptation to just build viral attention. “Sometimes you have to take a step back and realize that this thing that I’m doing is not being received in the same way that I imagine it,” he said during the interview.For students interested in entrepreneurship, his journey highlights an important lesson: successful startups often emerge from sustained experimentation rather than a single breakthrough idea.
Because I believe the next AI challenge isn’t intelligence – it’s memory
While much of today’s AI conversation revolves around increasingly powerful language models, Shah argues that the next big challenge lies elsewhere.I believe that in the future, each individual could have a personal AI agent, as people today have their own smartphones. In that world, he says, what distinguishes one agent from another will not only be the underlying model, but how well it remembers its use.“Everyone has their own AI agent just like everyone has their own phone,” says Shah. “In that world the most important thing will be the context for you.”He argues that context infrastructure—the systems that enable AI to remember long-term preferences, conversations, and information—will become as fundamental as the inference models that power today’s AI applications. Instead of treating memory as an add-on, developers should think of it as core infrastructure.
Because conviction mattered more than abandoning it
Unlike many startup stories, Shah says dropping out of college was never the goal. Describe enjoying college, doing well academically and making lifelong friends. The decision came only after spending nearly three years building and refining the technology behind Supermemory.His family encouraged him to complete his degree, and visa uncertainties added another layer of risk. Still, Shah says years of working deeply on the problem gave him the confidence to pursue it full-time.He also credits the Solo Founders program in San Francisco with shaping his growth, not because it encouraged founders to work alone, but because it surrounded them with peers who constantly challenged others’ thinking. Conversations about engineering, discipline, sales and business building have become part of everyday life.For students hoping to build careers in technology, Shah’s advice is to create and share work consistently instead of chasing online validation. He notes that many of his initial projects attracted little attention, but the publication helped him develop his skills, connect with future collaborators and eventually build credibility.As AI continues to reshape the industry, Shah’s story offers an alternative blueprint for aspiring founders: build iteratively, stay open to changing directions, and let curiosity, not hype, determine what comes next.
