Untitled business
lead
engineer
Overview
PlateLens is an AI nutrition coach that analyzes meal photos to estimate calories, protein, carbs, and fat, then provides personalized plate-swap suggestions to help users reach their fitness goals—whether getting lean or building muscle. Users snap a photo of their plate, get instant macro breakdowns powered by Google Gemini vision AI, and receive coaching on what to change to hit their targets. The app tracks meals over time and learns user patterns to give increasingly personalized advice.
Mission
PlateLens — AI Plate Coach
What we're building: An AI nutrition app that estimates the calories, protein, carbs and fat in a meal from a photo, tracks them for you, tells you what to change on your plate (swap this for that) to fit your goal, and coaches you toward getting lean or getting buff.
Vision backend — decided: Google Gemini (flash-lite class). One vision call per photo returns structured JSON of foods + macros; $0.0014/scan ($1.40 per 1,000 scans); OpenAI as drop-in runner-up; USDA FoodData Central as the canonical macro-lookup layer if LLM numbers drift. Privacy: process-then-delete photos, paid tier only (free tier trains on content). See /home/team/shared/vision-backend-recommendation.md for the full research.
Current focus — core app prototype: upload a photo → calories, protein, carbs & fat → goal-aware plate-fix suggestions (Lean/Buff) — built on the same published site.
MVP (landing page) — shipped: Hero ("Snap your plate. Know what you're eating."), how-it-works, goal teaser, early-access waitlist signup — live and verified.