Cal AI: a photo calorie counter that MyFitnessPal bought within two years

What it is

Take a picture of your food and the app logs calories and macros. It uses models from Anthropic and OpenAI plus retrieval to improve accuracy (TechCrunch, Mar 2025).

Timeline

• Before Cal AI: Yadegari built a games website in ninth grade and sold it for $100,000 at 16, by his account (TechCrunch).
• May 2024: launched. By March 2025 the founder claimed 5M+ downloads, 30%+ retention and $2M+ revenue the previous month; TechCrunch could not verify these but noted a 4.8-star App Store rating on 66,000 reviews (TechCrunch).
• Dec 2025: acquisition by MyFitnessPal closed after almost a year of talks (TechCrunch, Mar 2026).
• Mar 2026: MyFitnessPal said Cal AI had 15M+ downloads and $30M+ annual revenue; the 7-person team was retained and the app stays independent (TechCrunch).

Why it worked

• Simplicity over accuracy. MyFitnessPal's CEO: Cal AI serves people who "want it fast… They want it to not interfere with their life" (TechCrunch).
• Built natively on new models, rather than bolting AI onto an old app (TechCrunch).
• Distribution as a founding role. A co-founder, Jake Castillo, ran influencer marketing as COO (TechCrunch).

What to learn

1. "It already exists" isn't a verdict. Cal AI competed with MyFitnessPal's Meal Scan and SnapCalorie (TechCrunch) and won on speed and marketing.
2. Treat distribution (influencers, short video) as core work for consumer apps, not an afterthought.
3. Rising in public app rankings can make you an acquisition target: MyFitnessPal noticed it via Sensor Tower (TechCrunch).

Risks and what didn't work

• The early numbers were founder claims that TechCrunch could not validate (TechCrunch).
• Photo estimates trade accuracy for speed; the founders say about 90% accurate (TechCrunch).
• Consumer subscription apps usually depend on paid and influencer acquisition, which is expensive to replicate.