Tracking baby poop with AI
When our baby Arthur was born a week ago, my wife and I quickly realized a common issue: all baby tracking apps were designed for a single parent. We couldn’t find any app that allowed us to easily share information between both mom and dad. So, I decided to build one myself, coding directly from the hospital. What a better challenge to test the new GPT-o1 reasoning capabilities?
The Idea: A Baby Tracking App for Both Parents
My goal was simple: create an app that could track our newborn’s daily activities and be accessible to both parents. We needed to log feedings (whether breast or formula, and how many mL), diaper changes (wet or dirty), and have a real-time “timer” on the home screen showing when the last event occurred for each type (feeding, pee, poop).
Most importantly, I wanted it to be a smooth, user-friendly experience with an appealing interface—pastel blue tones, rounded buttons, and intuitive forms.
Using GPT-o1-Preview to Build the App
I decided to leverage GPT-o1-Preview to help me build the app from scratch using Streamlit, with local memory for data storage. The prompt I used to kickstart the project was:
“I need an app to track some activities of my newborn, specifically feedings, and whether they are breast or formula (if formula, also how many mL were consumed), and when he had a wet or dirty diaper. In the app, besides being able to easily add new events of this type and edit past ones, I want to always see on the home screen a ‘timer’ that tells me when the last event occurred for each type (feeding, pee, poop). I can only use Streamlit and local memory for data storage. I also want the GUI to be beautiful and engaging, with soft pastel blue colors and rounded buttons/forms.”
This prompt createad an app that was 99% perfect! The only adjustment and iteration was due to user experience, and personal feeling about having a “button” here or there, or changing the input method for pee and poop events from a dropdown to separate buttons.
Besides this few shot iteration the app was up and running in less than an hour, here you can see the final result:
The Finished Product
The app allows my wife and me to log feedings and diaper changes seamlessly, with the latest event information always at the forefront. It’s accessible, simple, and built to support both of us during those first intense weeks of parenting.
You can check out the code on my GitHub here. If you’re in a similar situation—or just curious about what you can build with AI—I hope this project inspires you.
For ease of use when we are “off-site” i then developed the app on a Azure WebApp service, to make ti avaliable from our mobile!
So far, take home message
This experience taught me that sometimes the best solutions come from a mix of personal necessity and a willingness to experiment with new tools. GPT-o1-Preview turned out to be an invaluable partner in this project, helping me create a functional, user-friendly app quickly and efficiently. Of course you can tell even this blog post was written, in between diaper changes, with the help of AI :)
I was impressed (and this bit is all mine) by the reasoning capabilities of GPT-o1 (preview) reasoning skills. It created the code, explained the choices, and made really smart adjustment in jsut the fist iteration! I was impress, but grateful, since we have very little time with a baby and a toddler, but i manage to be a father and a programmer straight away!
Have you ever built something out of pure necessity? Let me know in the comments—I’d love to hear your stories!