Why I'm building Nutrogine after Cal AI sold to MyFitnessPal

On March 2, 2026, MyFitnessPal announced its acquisition of Cal AI — the photo-scan calorie app built by Zach Yadegari and his high school friend Henry when they were 17. The deal had been negotiated for almost a year and closed in December 2025. Cal AI hit over 15 million downloads and $30M+ in revenue in under two years — a remarkable run for a high-school project.

That’s an absolutely deserved win. But it also closed a window I think the market still needs.

What changed when Cal AI joined MyFitnessPal

MyFitnessPal didn’t just buy the brand — it integrated Cal AI’s photo recognition into its own 20-million-food, 68,500-brand database that covers meals at 380+ restaurant chains. On paper that sounds like a strict upgrade: Cal AI’s UX, MyFitnessPal’s data depth.

In practice, it consolidates two of the four most-used calorie apps under one company, and it bakes Cal AI’s photo scanning into a database with roughly 20-25% entry error by random sampling (see our audit detail in the USDA vs MFP database analysis) — because most of MFP’s database is user-submitted with no systematic verification. Now Cal AI’s photo guesses are being matched against entries that themselves are guesses.

The gap this created

I’m not anti-Cal AI. Yadegari and his co-founder built something genuinely useful, very fast, and they got paid. Good. But three things bother me about where the post-acquisition market sits in May 2026:

1. Source attribution disappeared as a feature. None of the major calorie apps — Cal AI, MyFitnessPal, Cronometer, MacroFactor, Lose It, Noom — show you which source produced a given calorie number. You see “590 cal” and you trust it (or don’t), but you can’t click through to USDA, the brand’s nutrition page, or a customer report to verify. That kind of transparency is just absent from the category.

2. Restaurant context is still treated as a database problem, not a methodology problem. Adding more chains to a database doesn’t fix the fundamental issue: portions vary 14 to 27 ounces between Chipotle locations (Wells Fargo analyst Zachary Fadem and team weighed 75 identical bowls across 8 NYC stores in 2024 and confirmed it). Apps that show you a single confident number are wrong even when the number matches the brand’s spec sheet — because the brand’s spec is the idealized portion, not what’s in your bowl today.

3. The crowdsourcing trade-off is unspoken. MFP’s pitch is database scale. The unstated cost is that crowdsourced consumer nutrition databases — across MFP, Lose It, FatSecret, and similar — show entries averaging roughly 15-25% disagreement with USDA-verified or brand-published reference values across spot-check audits (see our USDA vs MFP database analysis for the underlying audit methodology). Most users don’t know that’s the trade-off they’re making.

Why I think this is solvable

I’m not a registered dietitian. I’m a product manager who started counting calories in 2024 and got annoyed at how much three different apps disagreed about a single Chipotle bowl. The angle that worked for me wasn’t picking a “best” app — it was reading Reddit threads about portion variance, looking up the actual USDA FoodData Central entries for individual ingredients, and cross-referencing what brands publish with what customers actually receive.

That research is what Nutrogine is. Every page on the site shows a calorie number, and next to that number there’s a Source Badge: USDA verified, Brand-claimed, User-reported, or Estimated. Clicking the badge takes you to the actual primary source — the USDA FoodData Central record, the brand’s nutrition PDF, or the specific Reddit thread.

It’s not a more complete database than MyFitnessPal. It’s a more honest one.

What I’m building toward

The site is the first half. The second half — coming Q3 2026 — is an app that does the same thing in your hand: photo scan, restaurant-aware lookup, cross-reference with USDA data, show you the source of every macro. No account required for core features. Free at launch.

If the consolidation makes you want to look beyond Cal AI / MyFitnessPal, that’s the gap I’m trying to fill. Join the waitlist below — I’ll email you when it ships, and roughly once a month between now and then with what I’m finding in the data.

FAQ

Is Cal AI still worth using after the MyFitnessPal acquisition?

Yes, if you’re already a happy user — Cal AI will remain an independent app per the acquisition terms, and Yadegari is still running the team. The integration with MFP’s database is the meaningful change: you get more food coverage, but the underlying entries are mostly crowdsourced.

How is Nutrogine different from Cal AI?

Three differences: every number on Nutrogine has a clickable source badge (USDA, brand, user-reported, or estimated); the site is restaurant-first, not home-cooking-first; and we don’t physically test food — we aggregate from public sources and cite each one.

When will the Nutrogine app launch?

Targeting Q3 2026 (July–September). Waitlist subscribers get access first.