Broccoli Meal Tracker

Track your nutrition effortlessly with AI-powered meal logging and analysis

Comment

Inspiration

I wanted to track my nutrition but all the apps I tried wanted me to scan through lists of ingredients and weigh things, which took ages. Either that or they scanned bar codes, which home cooked meals don't have.

I decided to make a simple app that could give me a rough idea of my nutrition quickly, using AI.

What it does

Broccoli takes pictures of your meals and passes them to an LLM which estimates their contents and nutrition. It stores the results in a database and gives the user a food diary so they can track calorie intake etc. throughout the week. They can set a daily calorie target too, and the app will show their progress against that using traffic light colours.

How I built it

I designed everything in Figma as a clickable prototype first. I coded the frontend using Expo / React Native / Typescript. I used Supabase for the backend, with Postgres functions for complex queries and Edge functions (with Deno) to talk to the LLM (Anthropic Claude) and parse the results back into the database. I used RevenueCat to implement a subscription pay wall after users have logged a set number of meals.

Challenges we ran into

Time! I'm a single developer with a full time job, leaving just evenings and weekends for this. It was fun though. The whole tech stack, RevenueCat SDK included, was a dream to work with.

Accomplishments that I'm proud of

  1. Making a full stack app, solo, that uses AI, which is "totally hot right now" so I've heard 😂
  2. Implementing in-app payments for the first time. In my day job, billing is managed away from our apps and I'd wanted to try an in-app integration forever.
  3. Making something that my friends are already using & enjoying (Yes, I had to enrol a load of testers for Play store submission, which is still pending)

What I learned

  1. RevenueCat docs & JS SDK are great
  2. Supabase, Postgrest & Tanstack Query are a really efficient way to make & integrate a backend
  3. AI models are better at estimating nutrition than me
  4. I need to eat healthier 😂

What's next for Broccoli Meal Tracker

Plenty! There's still some work needed on the meal editing functionality and diary UI. It also needs some social sharing options as a means of promotion. I also want to use more AI to add diet recommendations based on monthly data. That could be a good subscription incentive too 😄

Built With

Share this project:

Updates