Inspiration
I was playing around and found out that you can export all your Apple health data. I've been wearing an Apple watch for 8 years and whoop for 3 years. I always check my day to day and week to week stats but I never looked at the data over the years.
I exported my data and there was 1.2GB of data! Using A.I. to unlock data over 8 years gave me some insights I've never seen before.
What it does
It's a Mac application that takes in large data sets of Apple health data for analysis. It also supports calling cloud LLMs and local LLMs for analysis. It has a high privacy focus, so you don't need to send your data to the AI overlords.
How we built it
I vibe-coded the entire app. 1K views on youtube. Process here: https://www.youtube.com/watch?v=K036LsdfwgM.
Challenges we ran into
Turns out there are hundreds of types of Apple data. Creating visualizations for every single type and making sure there is no corrupt data is really hard. Also, processing gigabytes of data is actually harder than people think.
More importantly, if I provide AI functionality, it will cost me a lot of money that I can't predict.
Accomplishments that we're proud of
- 206 upvotes on Reddit: https://www.reddit.com/r/QuantifiedSelf/comments/1i7uu6c/i_open_sourced_my_project_to_analyze_your_years/
- $129 in sales since launch
- 250 stars on Github https://github.com/krumjahn/applehealth
What we learned
Using Revenue Cat to cover the AI Calls functionality to make sure we don't lose money on the app.
What's next for Health Data AI Analyzer
Our users have been asking us to make the process of downloading and analyzing the data automatic. We will offer a subscription service where you will periodically download your latest Apple health data and give you updates on your analysis. This will require an iPhone app and some sort of server to periodically process this.
Built With
- claude
- objective-c
- xcode
Log in or sign up for Devpost to join the conversation.