Blog
Notes from building Nutrogine
Posts from Alec on restaurant calorie data, AI accuracy, and the mess of crowdsourced nutrition databases.
-
CORNERSTONE
Cal AI's $50M MyFitnessPal exit: deal terms + what changes
MyFitnessPal acquired Cal AI in March 2026, closed December 2025. Founder-PM analysis of the mechanics, why MFP wanted it, and which alternatives still matter post-consolidation.
-
CORNERSTONE
Chipotle bowl weight: customer reports + Wells Fargo data
Chipotle bowl weights vary 14–27 oz across stores per Wells Fargo's 75-bowl test. Aggregating customer reports plus news coverage shows the same pattern: same recipe, real-world calorie load swings ±30%.
-
CORNERSTONE
Chipotle vs CAVA high-protein bowls: macro head-to-head (2026)
Chipotle's High Protein Menu (Dec 2025) vs CAVA's January 2026 menu launch. Macros, costs, and use-case verdicts — cutting, bulking, GLP-1, and value — for both chains' fitness-positioned bowls.
-
Best Cal AI alternatives after the MyFitnessPal acquisition (2026)
Cal AI's MFP acquisition consolidated the photo calorie market. Here are 6 alternatives ranked by what you actually need — accuracy, speed, free tier, or restaurant focus.
-
CORNERSTONE
The real accuracy of AI food-photo calorie counting in 2026
AI photo apps claim 92-97% accuracy. That's for simple foods. For real meals — stir-fries, restaurant bowls — accuracy collapses. What 2026 research really shows about Cal AI, SnapCalorie, and friends.
-
CORNERSTONE
USDA FoodData Central vs MyFitnessPal: honest database comparison
MyFitnessPal: 20M entries, mostly crowdsourced, ~23% error rate. USDA: ~13K entries, lab-verified, public domain. Why the smaller database is often the more honest answer.
-
CORNERSTONE
Why I'm building Nutrogine after Cal AI sold to MyFitnessPal
Cal AI's $50M ARR exit to MyFitnessPal consolidates the AI calorie market — but it also creates a gap for transparent, source-cited restaurant data. Here's the why behind Nutrogine.