How Cal AI Hit a $35M Run Rate in Year One, Then Sold to MyFitnessPal

One year after launch, Cal AI had not made $35 million. What it had done, on tape, is arguably more interesting. On May 15, 2025, co-founder and CEO Zach Yadegari told Nathan Latka that the photo-based calorie-tracking app had collected "a little over 10" million dollars since January — a figure he corrected upward, live, from the "9,000,000" he'd offered a few minutes earlier — and that based on its annual run rate the company expected "over 35 million" in the next twelve months. The first line of code was written in March 2024. The app launched that May. So the honest one-year scoreboard: $1 million in sales in the first five months, eight figures collected by month twelve, a $35 million run rate, zero dollars raised, 17 full-time people.

Cal AI revenue: the numbers on the record

Every figure here traces to the May 15, 2025 Latka interview or to Cal AI's GetLatka company page, which cites its sources row by row.

The exchange on the 2025 number is worth quoting, because it shows how loosely even founders hold their own topline. Asked what he'd collected year to date, Yadegari first landed on $9 million. Minutes later, unprompted: "I was backtracking on the revenue calculation, probably closer to like a little over 10 right now." Latka's on-air summary framed full-year 2025 as roughly $30 million; Yadegari pushed back only in one direction — "that's assuming no growth, but accounting for growth, it should be higher."

GetLatka's own estimate for full-year 2025 is a range of roughly $30–35 million, built from the stated run rate as the ceiling — the company never published a confirmed annual figure for 2025.

Does the arithmetic close? Cal AI's plan is $29 per year (displayed as $2.49 per month), so a $35 million run rate implies on the order of 1.2 million active annual subscriptions. Against 20,000–30,000 new downloads per day with 20–25% entering a paid plan or three-day trial — Yadegari's own conversion numbers — that's plausible, allowing for trial drop-off and refunds. Nothing in the tape contradicts it.

March 2024 to $1 million in five months

Cal AI started as three co-founders — Yadegari, CTO Henry, and Jake, who runs both COO and CMO duties — with a fourth, Blake Anderson, joining as the app took off. Yadegari is direct about what made the product possible: not proprietary AI, but timing. "It definitely is only possible now because of the breakthroughs with AI. However, that's mainly because of APIs like ChatGPT. We did not train our own model to do this."

The original plan wasn't even a single app. The holding company was built to spin up multiple apps powered by the founders' one unfair advantage — an influencer network posting daily. Cal AI simply outgrew the portfolio thesis: "By the time we had finished development on those other ideas, we realized it didn't make sense to spend any time there given calai's growth rate. It deserved all of our attention."

The growth engine: 250 influencers on flat retainers

As of May 2025, Cal AI was paying 250 influencers fixed monthly retainers to integrate the app into their TikTok and Instagram content. Total monthly spend: mid six figures. The largest single creator costs tens of thousands of dollars per month for four posts. Pricing each deal starts from the account's views, then engagement — "not all views are valued the same."

Latka asked the obvious question: why pay flat fees instead of commission on installs?

That would be the dream, but the reality is that there's too much competition out here with the brands, and they want to be guaranteed a payment. And if we're not gonna do it, another brand will.

Creative stays with the creators — "they are the professionals, they know what goes viral for their audience" — and Cal AI's job is only to make sure the product is integrated well. The channel is paired with Yadegari's own feed: a Twitter post about his college admissions drew 28 million views by his count, and the follow-up sharing his personal statement drew 40 million, which snowballed into television appearances. Less charmed: at the time of the interview his Instagram account was banned, which he attributed to a targeted campaign by an impersonator account.

A $29-a-year paywall, tested relentlessly

The monetization stack is a three-day free trial into a $29 annual plan, with 20–25% of users who finish onboarding converting to trial or paying outright. Paywalls run on Superwall — "100%, our conversion rate is much higher for using Superwall. It has allowed us to run so many tests" — while feature experiments run through Mixpanel, and only started in earnest around January 2025.

The onboarding lessons are the most transferable part of the tape:

• Moving the sign-in screen to the end of onboarding was the single biggest drop-off fix, "by far."
• Adding questions that have zero effect on the output — like which diet you follow — raised conversion, apparently because invested time makes users likelier to pay.
• Thank-you and encouragement screens lifted conversion; animating those same screens lifted it again.

Yadegari is candid that the testing culture is younger than the revenue: "We are behind on that front and have mainly operated through intuition." That includes App Store keywords, chosen with no ASO tooling — "mainly off the dome," as he put it, before Latka made him translate.

17 people, about half of them marketing

At taping, the team was 17 full-time, all remote, headed to roughly 22 within a month or two. About half worked on marketing, mostly managing the influencer program. Recruiting runs through X: two days before the interview, someone posted an unsolicited redesign of a Cal AI screen, and by taping he was interviewing for a full-time role. The hiring thesis is specialization — one person who only runs paywall experiments, one who only improves food-scanner accuracy, one who only owns the homepage. The stated number-one company value is speed.

Renewals were the open question. The app had just passed its first anniversary, so second-year data didn't exist yet; Yadegari's stated expectation was a renewal rate somewhere between 25 and 50 percent — a window he himself called "very broad."

What happened after the tape: the MyFitnessPal exit

The May 2025 projection landed, and then some. On March 2, 2026, MyFitnessPal announced it had acquired Cal AI — TechCrunch reported the deal had actually closed back in December 2025 after talks that ran the better part of a year, at a price neither side disclosed but which reporting characterized as high eight figures. In his own announcement on X, Yadegari wrote that he and Henry had started Cal AI as 17-year-old high-school students and that the app "broke $50m in ARR along the way"; at the announcement he put trailing-twelve-month sales at over $40 million — the figure Cal AI's GetLatka company page now carries for March 2026, citing the acquisition coverage on finance.yahoo.com.

That exit also resolves a number that would otherwise look strange: the GetLatka page lists headcount at 7 as of March 2026, down from the 17 full-timers on tape. Per the acquisition coverage, that seven-person team — Yadegari included — was retained by MyFitnessPal, with Cal AI continuing as a standalone product.

The clean version of the story, then: a four-person founding team wrote its first code in March 2024, hit $1 million in five months, collected over $10 million in the app's first twelve-ish months of monetization, exited year one at a $35 million run rate with no outside capital — and roughly ten months later sold the company to MyFitnessPal, having broken $50 million in ARR, in one of the fastest zero-to-exit runs consumer software has produced.