CalFit.AI & TrustNutr.AI — Dual AI Experiences from Scratch

Traditional calorie-counting and fitness apps suffer from acute drop-off rates because manual logging is tedious, repetitive, and mentally exhausting. Users are forced to search text databases, estimate portion gram weights, and record 6 to 10 separate ingredients for every single meal.

As Lead Mobile Product Designer, I partnered directly with the founder and ML engineering team to conceive, architect, and ship two complementary companion products: CalFit.AI, which replaces manual entry with real-time computer-vision meal scanning and macronutrient analysis, and TrustNutr.AI, an intelligent pantry copilot that scans refrigerator shelves to generate personalized, macro-aligned chef recipes.

The UX friction in mobile AI tools stems from the unpredictable latency and imperfection of machine-learning models. If a user points their camera at a complex salad and receives a slow 4-second loading spinner or inaccurate ingredient tags, trust evaporates immediately.

We had to design an interface that delivers instant perceived performance (under 400ms feedback loop), offers delightful micro-adjustments for portion confidence, and bridges nutrition tracking with practical recipe inspiration from whatever ingredients are already in the fridge.

We introduced an augmented reality camera overlay with fluid bounding boxes that identify ingredients dynamically as the camera sweeps over the plate. A single one-tap confirmation logs macros instantly to the user's daily goals with delightful haptic feedback.

Both applications launched simultaneously to the Apple App Store and Google Play, achieving a 4.9★ average rating across 12,000+ reviews and demonstrating an industry-leading 82% 30-day user retention rate.