wut is Cal AI?

How it scores

Transparency

Can you find out what it costs, who owns it, and how it works — before you pay?

Privacy

What happens to what you feed it, and who else gets to see it.

Value

Does the thing it charges for actually work well enough to be worth the money?

Staying power

Will this still exist, and still be the same product, in two years?

Wut it actually does

You photograph a meal, and the app identifies the visible food and estimates calories, protein, carbs and fat. It logs to a diary and can read and write to Apple HealthKit or Google Health Connect. That is the whole product, and its advantage over MyFitnessPal has always been speed — no searching a database, no weighing, just a photo.

It works. That is worth saying plainly, because the criticisms below are about accuracy and ownership, not about whether the thing functions. It reached over 15 million downloads and more than $30 million in annual revenue in under two years, which does not happen to an app that does not work.

Wut the research says about the numbers

This is the part worth reading carefully. Researchers at the National Institute of Diabetes and Digestive and Kidney Diseases, part of the NIH, photographed 102 meals prepared in a controlled metabolic kitchen with ingredients measured to 0.1-gram precision, then fed the photos to four apps: MyFitnessPal, LoseIt!, Cal AI and Appediet.

Across all four, calorie estimates averaged 250 to 345 calories per meal too low, and fat was underestimated by roughly 30 grams. Carbohydrates were the most consistently estimated macro. High-fat meals caused the most trouble.

Two honest caveats, because they matter. First, this was presented on 25 July 2026 as a conference abstract at NUTRITION 2026 — it is preliminary and has not completed peer review, so treat it as a strong signal rather than a settled fact. Second, the finding is about the whole category, not a Cal AI defect specifically; MyFitnessPal and LoseIt! were in the same study and had the same directional problem. No photo app has published peer-reviewed validation of a specific accuracy rate.

The direction is what should worry you more than the size. An error that runs low, on a tool people buy specifically to eat less, systematically tells you that you have more room left than you do. Three meals a day at the midpoint of that range is a meaningful daily gap.

Wut's the deal with who owns it

The story everyone heard is true: Cal AI was built by two teenagers, Zach Yadegari and Henry Langmack, while they were in high school. It is a good story and it sold a lot of downloads.

It is also no longer the situation. MyFitnessPal acquired Cal AI in a deal that closed in December 2025 and was only announced publicly on 2 March 2026 — roughly three months during which the app was marketed on an independence it no longer had. Terms were not disclosed. The seven-person team was retained, with no stated retention period.

Follow the chain up and it gets further from the bedroom-coder framing. MyFitnessPal was bought by Under Armour for $475 million in 2015, then sold to the private equity firm Francisco Partners for $345 million in 2020. So the scrappy teen app is now a private-equity-owned asset inside the incumbent it was disrupting. Cal AI still ships as its own app, and MyFitnessPal says it has no immediate plans to fold it into the main product.

None of that makes the app worse today. It does mean the thing you are choosing between — plucky independent versus corporate incumbent — is not a real choice anymore. And PE-owned consumer subscriptions have a well-known trajectory on pricing.

Wut it costs — and why you can't find out

You cannot see the price before you install. Cal AI does not publish pricing on its website or its App Store listing; the cost appears only after you complete an onboarding flow inside the app. That is a deliberate design, and it is the single strongest mark against it on transparency.

The App Store listing does expose the in-app purchase price points, which is the closest thing to a public price list: entries named "Unlimited" appear at $2.99, $5.99, $9.99, $19.99 and $29.99, plus an "Unlimited Plan" at $19.99 and a "Streak Restore" at $0.99.

Five different price points for a product called "Unlimited" is not a tiering structure — it is price testing. Which means the price you are shown is not necessarily the price someone else is shown, and you have no way to know where in that range you landed.

Wut happens to your food photos

Here is the gap. Cal AI's privacy notice, effective 21 May 2026, confirms your food images are sent to its API and service providers for analysis, and that it stores the resulting food names, calorie and macro estimates, timestamps and metadata. Fine — that is the product working.

What the policy does not say is whether your photos are used to train or improve its models, or how long the images themselves are retained. It refers generally to "machine learning systems, and artificial intelligence" for service enhancement and commits only to keeping data "as long as necessary." For an app whose entire moat is a food-recognition model, silence on whether your photos feed that model is a conspicuous omission rather than a technicality.

The sharper detail is what is protected and what isn't. Data pulled from Apple HealthKit is walled off from advertising — Cal AI states HealthKit data is not used for marketing or transferred to third parties for it. But that protection exists because Apple mandates it, and it covers HealthKit data specifically. The food and activity data you log directly in the app is a different bucket, and the policy lists sharing with "social media platforms, third-party advertising networks" and the use of targeted advertising cookies. The App Store privacy label backs this up: it declares identifiers and usage data are used to track you across other companies' apps and websites, and lists third-party advertising as a purpose.

The policy also notes food data "may include sensitive personal information when the information indicates or allows someone to infer a health condition." You can request deletion at privacy@calai.app.

Will it still be here in two years

Yes, more likely than almost anything else in this category — and that is the one thing the acquisition genuinely improved. A viral app owned by two 19-year-olds is a wrapper risk; the same app inside MyFitnessPal, with a nutrition database of 20 million foods and 68,500 brands behind it, is not going to vanish because a founder gets bored.

The risk shifts rather than disappears. What changes under private-equity ownership is usually the price and the free tier, not the servers. Expect the product to persist and the terms to get worse.

Sources

Facts last checked against these on 2026-07-28.

1. NIH / NIDDK study presented at NUTRITION 2026 (ScienceDaily summary)102 metabolic-kitchen meals; apps under-estimated by 250–345 kcal and ~30g fat; Cal AI among the four tested; presented 25 Jul 2026, not yet peer reviewed.
2. TechCrunch — MyFitnessPal has acquired Cal AIDeal closed Dec 2025, announced 2 Mar 2026; terms undisclosed; 15M+ downloads, $30M+ annual revenue; founders Yadegari and Langmack; 7 staff retained.
3. Under Armour — completion of MyFitnessPal sale to Francisco PartnersMyFitnessPal sold to private equity firm Francisco Partners for $345M, completed Dec 2020; Under Armour had paid $475M in 2015.
4. Cal AI privacy notice (effective 21 May 2026)Image handling; no stated photo retention period or model-training disclosure; sharing with advertising networks; HealthKit advertising carve-out; deletion contact.
5. Cal AI on the App StoreIn-app purchase price points ($0.99–$29.99); privacy label declaring cross-app tracking and third-party advertising; 4.8 stars from ~347,000 ratings.