He Sold a Company at 16. Raised $3M at 19. Then Gave His AI Engine Away for Free.

Key Takeaways

• He got the résumé backwards — on purpose. Sold his first company at 16. Raised $3M at 19 as a solo founder. His open-source engine now has 26K+ GitHub stars. The whole point: he got there by not chasing any of it — and the day this episode drops, he open-sourced the engine so anyone can run it locally, for free.
• The raise is a result, not a goal. Dhravya said no to VCs — including a16z — for nine months, while broke and borrowing from friends. He only raised once the company was undeniable. The $3M came after the work, not before it.
• Build your art in public until it becomes a company. He never tried to find a startup idea. He made things he was obsessed with — bots, a viral database, a meme tool — open-sourced all of it, charged for none of it. One of them (Any Context) became Supermemory.
• Kill your own viral hit. A product he launched got 500K impressions in a day, mid-fundraise. He buried it anyway, because it wasn't the future. Iteration velocity beats validation.
• Solo, after a co-founder nearly destroyed him. An earlier company died in a co-founder breakup — sold "minutes before they destroyed everything." He doesn't want to argue about direction every morning.
• The honest version of AI memory. Why the category is full of benchmark-gaming, why ChatGPT and Claude "fail" memory benchmarks on purpose, and what evals actually matter.

Listen: YouTube | Apple Podcasts | Spotify

Introduction

Dhravya Shah sold his first company at 16. At 19, he raised $3M as a solo founder for Supermemory — the memory and context layer for AI agents, with an open-source repo now past 26K+ GitHub stars. And the day this episode goes live, he's doing the most on-brand thing imaginable: open-sourcing the whole engine so anyone can run it locally, for free. He's the youngest, earliest-stage guest we've had on the show — and the most interesting thing about that résumé is that he never chased any of it.

He's a college dropout from Mumbai by way of Arizona State, and that's exactly why Julian wanted him on. Almost everyone else we talk to has years of distance from the moment they were figuring it out. Dhravya is in it, in real time. His path to a venture-backed company is the cleanest argument I've heard against the way most people try to start one.

The short version: he spent years building things he loved, in public, for free, never calling any of it a startup. One of those experiments blew up to 100,000 users while he was in negative bank balance, borrowing money from friends to survive. VCs chased him for nine months — he turned them all down, including a16z — until the company crystallized into something he couldn't not build. Only then did he raise. In this conversation with Julian, he makes the case that the fundraise was a result of all of that, never the point of it.

The Raise Is a Result, Not a Goal

When Any Context — his open-source tool for collecting personal context — hit thousands of GitHub stars in a few days, the VCs found him fast. He took the meetings, mostly out of curiosity, and kept giving them the same answer.

"You're telling me I can raise $2 million right now? I don't think that's the case. It was never even a question or a goal for me. The goal was to build the version of the world I thought had to be true."

He said no for nine months, from the summer of 2024 to the summer of 2025. The discipline turned out to matter enormously. Back then he was building a consumer app — if he'd raised on that thesis, he'd have been locked into consumer investors right before he needed to pivot the entire company toward infrastructure. "It's hard to turn a big ship," he said. Staying small and saying no is what let him raise, later, from exactly the right people.

The framing that stuck with him came from Dane Knecht, Cloudflare's CTO and an early mentor: raise from a tier-one investor, bootstrap, or "make something that makes so much sense you don't even have to ask the question." Raising before that point of obviousness, Dhravya argues, makes everything harder — now you have to figure out the company and you don't even know who to raise from. The money should be a consequence of an undeniable thing, not the thing that's supposed to make it undeniable.

Build Your Art in Public Until It Becomes a Company

The reason this works for Dhravya — and the reason it's so repeatable — is that he never optimized for the outcome. He built in basically every category: a database that went viral, a meme generator, automation bots, two companies he sold as a teenager in India.

"I never approached them as startup ideas. I approached them as something cool I made. It was my way of doing art."

Everything was open source. Nothing was behind a paywall. And the audience that accrued — people who followed him because he was visibly in it for the love of the game — is what made Any Context land when it finally arrived. This is also his answer to the most common excuse he hears, the one about not having connections. He was a kid in Mumbai, then a kid at ASU, a school nobody associates with startups. The network came entirely from putting work into the world. It's a pattern he still hasn't broken: the day this episode airs, he open-sourced Supermemory's core engine — free to run locally on any machine — instead of locking it behind a price.

His advice on posting is almost the opposite of growth-hacking:

"Your goal for posting shouldn't be for someone to look at it. It should be to show your real self on the internet. The network effect comes later — but you shouldn't expect it."

It won't get you followers for the first few years, he says, and that shouldn't be the goal anyway. Julian — who went to school for music — drew the parallel to playing a room with no one in it: if an empty room is enough to make you quit, you were never going to get to the full one.

Kill Your Own Viral Hit

The hardest discipline in the episode isn't saying no to VCs. It's saying no to your own success.

The day Dhravya moved into the Solo Founders Program house, he launched a router that got 500K organic impressions in a single day — in the middle of his fundraise. The validation was enormous. He buried the product anyway.

"If something's blowing up but it's not the future, I'll end it. If something is the future, I'll build it — even if it doesn't get the visual pop."

This is the inventor's dilemma, and Dhravya talks about it with unusual clarity for someone in the thick of it. The trap isn't failing to get traction — it's mistaking traction for direction. He reaches for the IBM System/360 story: the founder's son killed the company's profitable custom-machine business for two years to build the thing that would actually define the future. Most people, handed a viral hit mid-fundraise, would lock in. Dhravya's instinct is to ask whether it's the future first, and walk away if the answer is no.

Why Solo

Dhravya isn't solo by ideology. He's solo because he's lived the alternative. An earlier hosting company died in a co-founder breakup — one co-founder "went rogue," and they had to sell the business "minutes before they destroyed everything." He's trialed co-founders since, and kept running into the same walls: compatibility, different ways of working, and above all, directionality.

"I don't want to wake up every day and argue with someone about what we should do next. There's a lot of easy money everywhere — I want the thing that's hard, that takes years, where we might die."

That last part is the whole point, and it's where the episode lands its bull case. Solo founding, for Dhravya, is the only way to build something genuinely true to yourself — the kind of conviction you need to keep going in the dark, when there's no one beside you and no external reason to continue.

"Every company a solo founder builds is literally their representation, inside a company. When I think about who's building what, it makes complete sense that they're the one."

Solo doesn't mean alone, though. He's got a core team of seven, all with him for about a year, most recruited straight out of the open-source community — "true builders" doing the work for free before there was ever a job. And the Solo Founders Program house gave him something a co-founder never would: a dozen other people building solo down the hall, who hold him accountable without him having to hand over the wheel.

The Honest Version of AI Memory

For the builders in the audience, the back half of the conversation is a clinic on a hype-saturated category. Memory has the lowest barrier to entry in AI infrastructure, Dhravya argues — which is exactly why it attracts the most noise and the most benchmark-gaming.

"The best memory products — ChatGPT, Claude — actually perform badly on these benchmarks. Because they're built for the user, not for showing numbers on Twitter."

He invokes Goodhart's Law — when a measure becomes a target, it stops being a good measure — and walks through how a recent benchmark optimized for wording rather than long-term recall. His own answer is Memory Bench, a tool for running evals in the most deterministic environment you can get, reporting latency, quality, recall, and tokens honestly, on the customer's own data. The product philosophy underneath it is what he calls "Lego blocks with no instructions": give developers configurable pieces — memory, retrieval, file systems, hybrid modes — and let them decide what's perfect for their use case, rather than hard-coding an opinion and forcing everyone into it.

About Dhravya Shah

Dhravya Shah is the founder and CEO of Supermemory, the universal memory and context layer for AI applications and agents — whose open-source repo has crossed 26K+ GitHub stars. Originally from Mumbai, he sold his first company at 16 (and a second as a teenager in India), came to Arizona State University as a self-described "lost student," and worked on open-source AI at Cloudflare — where the CTO became an early mentor — before dropping out to build Supermemory full-time. At 19 he raised $3M (a $2.6M seed led by Susa Ventures, per TechCrunch) from backers including Google's chief scientist Jeff Dean, Cloudflare CTO Dane Knecht, and Logan Kilpatrick. He's a solo founder and a member of the Solo Founders Program, with a core team of seven.