mymir vs Cal AI: Photo Logging Is a Feature, Not a Strategy

Cal AI made photo-based calorie estimation mainstream — famously built by two teenagers, it grew to over 15 million downloads and reportedly over $30 million in annual revenue before MyFitnessPal acquired it in December 2025. Point your camera at a plate, get a calorie and macro estimate back in seconds. That is a genuinely useful shortcut over typing every ingredient into a search bar, and it's one of the reasons photo logging has become table stakes for any new food app — mymir included.

The difference is what happens after the photo. Cal AI's job ends when the estimate lands in your daily total. mymir treats the photo as the start of the loop, not the end of it.

Same input, different output

Both apps take a picture and identify what is on the plate. From there:

• Cal AI logs the estimate and adds it to your daily calorie count, now backed by MyFitnessPal's database of roughly 20 million foods since the acquisition.
• mymir logs the estimate, then folds it into pattern detection (is this a repeat of a late-night habit, a low-fibre day, a sign of weekend drift?) and into whatever five-day experiment you have running — so the same photo either confirms or challenges something you are actually trying to change.

Now part of a bigger company — same underlying job

The acquisition matters for one reason: it's MyFitnessPal, the largest incumbent in food tracking, publicly deciding that AI photo logging was worth buying rather than building in-house. It doesn't change what Cal AI does day to day. It's still a fast, accurate way to log a meal — full stop. It still won't tell you whether the change you made last week is working.

Logging accuracy is necessary, not sufficient

A precise calorie estimate from a photo is a genuinely hard problem, and getting it right matters. But accuracy on a single meal does not answer the question most people actually want answered, which is closer to: "is what I'm doing this week working?" That requires looking across days, not just within one plate — which is why mymir pairs photo logging with a health score, weekly insights, and structured experiments rather than treating the estimate as the finished product.

Where mymir goes further

mymir also accepts plain text ("steak sandwich for dinner") and barcode scans, so photo logging is one option among three rather than the only path in. And when you go quiet for a day, mymir's coach checks in — not to guilt you about a broken streak, but because a gap in the data is itself information worth asking about.

Frequently asked