Fitineary
A fitness diary where your AI coach and your human coach share one screen, and logging a meal takes a photo.
The short version.
Fitineary is one diary for food, exercise, water, sleep and mood, with a coaching layer built in. Users log a meal by taking a photo, speaking, or typing, including in Malayalam and Manglish, and review the AI's estimate before anything is saved. A personal trainer or nutritionist can connect to a client, assign routines and see how closely they're being followed.
We took it from a clickable prototype to live apps on Google Play and the App Store in 38 days, then kept a weekly release rhythm driven by the client's own feedback.
What was broken.
People log food in one app and workouts in another, then hand their phone to their trainer to scroll through both. Calorie apps don't recognise puttu, appam or a plate of curd rice, typing every meal is tedious, and none of them give the trainer a real place in the app.
- Western food databases miss everyday Indian meals.
- Manual logging is slow, so people stop after a week.
- Trainers coach from screenshots and WhatsApp messages.
From idea to production.
- 5–6 Aug
Prototype signed off
A complete clickable prototype defined every screen and design token before any app code was written.
- 22 Aug
Version 1.0
The first full release of the app on a real backend.
- 12 Sep
Live on both stores
Version 1.3 went live on Google Play and the App Store in India.
- 23 Sep
Premium and feedback
First paid subscriptions, and a 35-point feedback round from the client turned into the 1.8 release within days.
What we built.
One Coach tab shows either the AI coach or the human coach, and the AI steps back in exactly the areas the human coach covers: training, nutrition or both. A routine is the same thing whether the user built it, the AI suggested it or a coach assigned it. Logging is designed to take seconds: point the camera at the plate, check the estimate, save. The app records everything on the phone first and syncs later, so it works in a basement gym with no signal.
- Step 1Snap a photo or speak
- Step 2AI estimates food and portions
- Step 3User reviews and saves
- Step 4Coach sees adherence
- Step 5Routine adjusted
What we delivered.
Photo, voice and text logging
Snap a plate, say what you ate, or type it. Malayalam and Manglish are understood.
AI coach and human coach
One Coach tab. The AI yields the areas your human coach takes on.
Routines that travel
The same routine whether you built it, the AI suggested it, or a coach assigned it.
Works offline
Logs save on the phone and sync when a connection returns.
Coach chat
Photo and voice messages, exercise demo videos and reminders.
Sign in with WhatsApp
A one-time code on WhatsApp or email, no passwords.
Full diary
Food, exercise, water, sleep and mood in one timeline.
Premium
Monthly to yearly plans at the same price on Android and iOS.
A closer look.
Intelligent, and trustworthy.
Model chosen by experiment
We tested 10 AI models on 28 real meal photos, over a third of them Indian dishes. The cheaper option over-counted by more than 200 calories, so we ruled it out.
Two cents per user
Photo and voice logging cost about $0.018 per user per month.
You stay in charge
Every AI estimate opens in the normal review screen. Nothing is saved until the user taps Save.
Private by default
The app never holds an AI key, requests are rate-limited, and the AI provider is required to keep no data.
The stack.
- Apps
- React Native (Expo) for Android and iOS
- Backend
- Supabase Postgres with row-level security, edge functions
- AI
- Gemini Flash via a secured Cloudflare Worker
- Business
- RevenueCat and Apple in-app purchase, Dodo Payments, Firebase crash reporting and push
The numbers.
Build proof
Expected impact for users
Verified from project repositories, test runs and release records. Projected modelled estimates of user impact, not measured client results.
Got an idea? Let's ship it.
Tell us the workflow that's costing you customers or hours. We'll come back with a prototype plan, a timeline in weeks, and a fixed quote.