Zach Yadegari didn't stumble into success. At 16, he sold a website for roughly $100,000. Two years later, taking what he had learned from his first project, he co-founded Cal AI, which went on to become a multimillion-dollar consumer app business with more than 15 million downloads and more than $30 million in annual revenue — all before his 20th birthday.

The interesting part of his story is not that he started as a young teen coding prodigy or that he moved quickly. Plenty of founders do both. What stands out is how deliberately his strategy evolved from one venture to the next: Each venture appears to have taught him something about finding demand, reaching customers and scaling what works.

His first major experiment was called Totally Science, a gaming website created during the pandemic. Schools had issued students Chromebooks while networks simultaneously blocked many gaming sites. Sitting in class, Yadegari noticed students had multiple tabs open with different games across different websites.

He saw an opportunity to combine them into one place: “I decided it would be much better if there was one website that was a hub for all of these games to play on.” He thought he could easily take the games already available elsewhere and put them on one site, creating something he knew students would use. At this point, he wasn't trying to build a company or even make money from it. His motivation was much simpler: he wanted people in his school using something he had built himself.

Totally Science did not require sophisticated technology; its advantage was understanding firsthand the environment in which the customer was operating. The site spread through students and later through social media, eventually drawing more than five million users during COVID-19.

At 16, Yadegari monetised the traffic with advertising and sold the business to FreezeNova for around $100,000. Rather than treating the exit as the conclusion, he used the capital to finance his next experiments, later saying he invested about $130,000 into Cal AI. The first venture had given him something beyond capital: an early lesson in finding demand that already exists and removing friction between the customer and that demand.

His next project, Cal AI, represented a significant upgrade to that approach. The problem was again one Yadegari had personally experienced: he wanted to track his food while trying to gain weight but found conventional calorie tracking troublesome and tedious. Instead of building another comprehensive nutrition platform, the team reduced the proposition to a simple question: “What if we could take a photograph of your food and receive an estimate of its calories and macronutrients?”

Cal AI’s business model was unusually compatible with social media distribution and marked a significant shift from Totally Science. This time around, distribution was no longer something to discover after building the product but something that had to be considered during product design.

The team then treated content itself as an experimental system, testing hundreds of hooks and formats and reaching out to creators rather than assuming they knew what would go viral. Yadegari initially made TikToks himself before expanding through influencers, particularly in health and fitness. The company used direct messages and emails to reach creators, starting with smaller campaigns before committing to larger deals, then measured performance and retained creators who produced results.

By 2025, Cal AI had turned influencer marketing into a serious acquisition machine. Yadegari said the company was paying 250 influencers on retainer, with the largest creators earning “tens of thousands per month” for four posts. Rather than dictate the content, Cal AI looked at each creator’s views, engagement and audience before deciding what they were worth, while leaving the actual creative concept to the creator: “They are the professionals. They know what goes viral for their audience.” The team also built out its own marketing operation to manage the growing network, with roughly half of its 17-person team focused on marketing and influencer management.

The same test-and-measure mentality eventually moved inside the product. Yadegari described speed as Cal AI’s “number one core value”: “push things out fast, develop it fast, push it out fast, see the test results quickly.” The team began systematically testing onboarding and paywalls rather than relying purely on intuition, using tools such as Superwall to run hundreds of paywall experiments. His thinking was simple: “We treat paywalls the same way we treat the product; we ship something, measure it, and improve it tomorrow.” At this point, Cal AI was no longer just an app with successful influencer marketing around it. It was becoming a machine for continuously testing product, conversion and distribution.

And the numbers show what that system produced. By May 2025, Yadegari reported an annual run rate of more than $35 million, 20,000 – 30,000 new downloads a day and 250 influencers on retainer, while the company remained completely bootstrapped. MyFitnessPal had been watching Cal AI climb the App Store rankings and eventually acquired the company in a deal that closed in December 2025 and was announced publicly in March 2026.

At the time of the announcement, MyFitnessPal said Cal AI had surpassed 15 million downloads and $30 million in annual revenue, with the seven-person Cal AI team retained and the app continuing to operate independently. What started as a simple calorie-tracking idea had become a consumer software business built around rapid experimentation and a distribution system that could be scaled.

Six months after the acquisition, Yadegari announced his newest venture, Persona, an AI assistant designed to live on your wrist. The September 2026 launch is a sharp pivot from calorie tracking into AI hardware: a $179 screenless aluminium wristband paired with an iPhone app, designed to let users handle emails, calls, calendars, purchases and other tasks through voice commands.

With Totally Science, Yadegari spotted an existing demand and built the simplest solution around it. With Cal AI, he took that one step further, designing the product with distribution in mind and then building a system around what worked. Persona is now testing whether that same approach can move into an entirely new category: AI agents and hardware.

The takeaways from his first two ventures are simple:

  • Start with a problem you understand. Build the simplest thing that solves it.

  • Get it in front of people quickly.

  • Test the distribution as aggressively as the product. Then turn whatever works into a system.

Yadegari's real asset is therefore not an individual app but the process he has accumulated across ventures: identifying consumer demand, building quickly, testing relentlessly, measuring what works and scaling the winners. That is a different venture philosophy from raising capital, building a large product and then searching for customers. His approach starts much closer to the customer — and increasingly, much closer to the distribution channel.

That makes Persona the most interesting test yet. If it works, Yadegari will have taken the same underlying playbook from a teenage gaming website, to a viral consumer app, to AI hardware in just a few years. Yadegari himself has talked about the possibility of building Persona into a company worth “billions,” though regardless of whether the venture reaches that scale, one fact remains: Yadegari's playbook is worth keeping an eye on.