VoCal: The world's easiest calorie tracking app

Role

  1. Founder, UX , full stack builder

  2. Designer, Developer(agentic coder with claude code)

  3. Self initiated

Methods

Beta testing, rapid prototyping, AI assisted design and code

Challenge

Building a voice based calorie tracking app which works for Indians like no other app.

VoCal

VoCal

The world’s easiest calorie tracking app.

VoCal is a voice-first nutrition tracking app that lets users log meals simply by speaking.


Instead of searching foods, weighing portions, or manually entering calories, users can say what they ate and VoCal converts it into calories, macros, reports, streaks, and personalized insights.

Ya i know!

There are already too many calorie tracking apps!!

BUT

The issue I faced is that I never eat in one sitting so taking a photo of my dish and then keeping refilling it and then taking pictures was not ideal and databases are even worse.

I wanted the resistance to tracking be almost null.

so I just built it

I am a product designer and not a coder but AI truly unlocked a lot of things for me and what followed was:

One week of design sprint in Figma.

Yes, the file was messy because before I could sort it out, I was already building it.

Figma MCP plus Claude Code

# CLAUDE.md

This file provides guidance to Claude Code (claude.ai/code) when working with code in this repository.

## Project Overview

VoCal is a voice-based calorie tracking mobile app built with React Native (Expo) that uses AI to process voice recordings into structured food data. Users speak their food intake, which is processed by Supabase Edge Functions using Sarvam.ai STT and LLM services.

**Core Technologies:**
- **Frontend**: React Native (Expo SDK ~53), TypeScript
- **UI Library**: Gluestack UI v3 with NativeWind (Tailwind CSS)
- **Backend**: Supabase (Auth, Database, Edge Functions, Storage)
- **Voice Processing**: expo-audio for recording
- **Speech-to-Text**: Sarvam.ai Saarika API
- **LLM Processing**: Gemini API (via Edge Functions)
- **Analytics**: PostHog
- **Payments**: RevenueCat
- **Node Version**: 22.16.0 (local dev); eas.json and package.json engines still reference 20.18.1 (EAS cloud builds use that)

Kept building, kept iterating, and the first version was out after 3 weeks into beta.

Beta testing

I kept documenting my journey on my Instagram, grew my Instagram to 20,000 followers, and had hundreds of beta testers who were posting their feedback, letting me know what is wrong, what is right, And posting it online helped get feedback from even users, and not only the beta testers.

Launch

Building was not as difficult as launching because the app got rejected two times.
I had to:

  • do the ASO(app store optimization)

  • write descriptions

  • make app listing images

  • comply with the rules

  • set up payment gateway, revenue tracking,

  • analytics

Post launch

  1. Set up attribution on AppsFlyer

  2. Started doing marketing and distribution because without that the app meant nothing

  3. Continued iterating on the basis of the comments, feedback, reviews

VoCal Features

  • Voice logging

    You just speak your meal in any of the 11 Indian languages or English and it logs it for you.

  • Analytics
    A place to see your trend progress numbers and track your weight loss, gain, or maintain

  • Reports

    A feature which no other app in the world has: customized weekly and monthly reports to tell you what exactly to eat, what not to eat, what is going wrong, what is right.

  • Streaks

    Streaks to keep you on track and a leaderboard to motivate you!

This app is as much of Claude code as it is mine :)

Impact

  1. 3,000 downloads, 300 paying customers in 2 months.

  2. Average 4.5 stars across both app stores.

  1. 3,000 downloads, 300 paying customers in 2 months.

  2. Average 4.5 stars across both app stores.

Learnings!

Learnings!

Challenge

Making a fully fledged app without knowing the technical constraints, without knowing how to code, but still making sure it works just as well and even better than every app out there

Surprise

AI is pretty good.

Insight

When designing and building it yourself, you find out about technical constraints, edge cases, states which you might have not even considered while designing, and simple things which get overlooked because of the disconnect between engineering and design.

Skills learnt

  • Agentic engineering, vibe coding

  • Fundamentals of MCP AI-assisted development and basically building

  • App store optimization

  • Attribution and analytics via MMPs and tools like post hog.

Create a free website with Framer, the website builder loved by startups, designers and agencies.