When my daughter was born, we started using Nara Baby so we could make sure she wasn’t eating too infrequent, sleeping too much or little, etc. I liked this tracker because it allowed us to turn on or off things we didn’t want to track (like solids as she was very little, etc). It also had the main feature we loved – a timer for sleep. One button to press to start, and then anyone with the app (my wife and I) could stop it on our phones when we swapped duties.
The app was great, and it also had amazing visualizations. All things must come to an end.
They recently announced that it was going paid only – and gave us a trial but we can still export data. So I did the only logical thing a software engineer handed 2.5mb of data would do.
I built a visualizer from the data.
I haven’t been using LLMs as much but one thing they are wonderful at, is analyzing data and picking apart formats. I started writing a PLAN.md file, and working with Claude CLI I started building my graphs.
I think my initial plan and prompts were “In this directory is a csv export from Nara Baby about my child. I want to keep track of all the events and measurements in a dashboard. Using charts, and tabler, show me the data”. Then over the day, I plucked away at more things to enhance.
This ended up with me making
https://babygraph.bureauofsidequests.com/
and being extremely happy with this project I made for myself.
I’m not sure how the rest of the world will feel about this, as I get what I want out of it.
The data in this blog post is horribly contrived, from the fake fixtures, but with real data you can see trends. It was SO cool seeing the Feedings by time of Day, overcoming some issues, and then finally where we started using bottles for bedtime and a nice trend line started appearing in the chaos.
