Zach Yadegari was a high school junior in Roslyn, New York when he and Henry Langmack started building a calorie tracker that worked from a photo. The two had met at a coding camp when they were ten and had shipped a run of apps together before this one. By Zach's own count, five of them had already launched and gone nowhere. He has been blunt about that record: "I launched 5 apps before Cal AI and none of them really took off. But I believed in myself enough to keep making more."

The money that started Cal AI came from the app before the apps. At 16 he sold Totally Science, an unblocked gaming site that had reached five million users, for $100,000, and he says he put all of it into Cal AI. Blake Anderson, whom he first reached by cold DM on X, put in a few hundred thousand more and brought the short form video playbook the company would later run on. Henry took the CTO seat. Jake Castillo joined later as the fourth co-founder. There was no outside funding at any point.

The timing anchor for everything below is his own post: Zach introduced Cal AI publicly on April 10, 2024, with a Product Hunt link and a buildinpublic tag. What he introduced did exactly one thing. "We launched Cal AI with a single feature," he wrote later. Point a camera at a plate, get calories and macros back, skip the weighing and the typing that had made him quit MyFitnessPal three days in.

Ninety eight days after that post, on July 17, 2024, he published the number that makes this story worth reading: "In less than 100 days, I have reached nearly $300k in monthly revenue on one app." An app running at roughly $300,000 a month had crossed $100,000 in annual recurring revenue many weeks before that, so read 98 days as a ceiling on the milestone rather than the exact day the line was passed. Even as a ceiling it is close to the fastest start in this database.

The curve did not flatten. On August 31, 2024 he wrote that the app had gone from zero to $500,000 MRR in four months. On September 5, the first day of his senior year of high school, he posted that Cal AI had passed $1 million in total revenue. By late November he was describing $1 million MRR at 17. In February 2025 the line was "We built Cal AI to $1.8 million MRR from our high school classrooms." In March 2025 he summarized it as zero to $24 million ARR in ten months, and on April 1, 2025, in the post about being rejected by 15 of the 18 colleges he applied to, he listed himself as 18 years old, 34 ACT, 4.0 GPA, $30M ARR biz.

What drove that was distribution, not the model. Cal AI ran on influencer video at volume with paid social behind it, and Zach has said plainly that "influencers is definitely the real sauce for Cal AI" while paid ads came later. His hiring posts show what the growth side was actually optimizing. One asks for a video editor because "the more videos you test, the more you win," and describes thousands of raw influencer clips waiting to be cut. Others ask for people who know Superwall, Mixpanel and Customer.io, which is paywall and lifecycle work rather than acquisition work. The clearest single number he has given on that effort is revenue per download: about $1 on average at launch, about $4 fourteen months later, which he credits to A/B testing.

The engineering stayed small and ordinary for an AI product. His job posts name TypeScript and microservices, PostgreSQL, Firebase, vector databases and SwiftUI on the client. The team reached 30 plus people, remote.

He is also public about the misses. In late September 2024 he wrote that he had lost roughly $30,000 by staying in the weeds on tasks he could have handed off instead of running a paywall test he had been sitting on for two weeks. When he finally ran it, it added around $2,000 a day.

On March 2, 2026 he posted that Cal AI had been acquired by MyFitnessPal, the incumbent whose manual food logging had annoyed him into building the thing. His framing: "In just 18 months, we've helped millions of people lose millions of pounds. And we broke $50m in ARR along the way." Cal AI continues as a separate app. In the same thread he named the original target: "Our North Star goal when starting the company was just to hit $50k / mo, and even that I felt was ambitious."

One caveat is worth keeping in view. Every figure here is self-reported on X and none of it was audited, and Zach's own numbers move. The acquisition post says $50 million ARR. Two months later he wrote "At 18 I sold Cal AI while at $40M ARR." In August 2026 he described it as "$50M in the first 18 months," which is cumulative revenue rather than a run rate. He did report $5.7 million for January 2026 on its own, which supports the higher end. Treat the range as the fact, not any single post.