Devin Jeffery

I call it raccoon mode — when I catch myself rummaging through my kitchen without a thought in my brain.

Raccoon mode kicks in automatically when I'm deep in thought working through something, after a meeting marathon, when I want to celebrate, or when I'm stressed. All that to say, it's not going anywhere.

Calorie tracking apps don’t work for me

Even the AI ones. If I go even 1 calorie over my goal, it feels like I failed. If I stayed under the calorie goal but my macros were off, it feels like I failed. I know these tools help a lot of people, they just don’t work for me.

I decided to build an app for myself

What I am looking for is balance and more mindful eating. This is where I had an idea: why don’t I use Cursor to build myself a food journal app? Some of these exist already but I wanted something simpler.

My idea was that before I’m about to eat something, I snap a quick picture of it and mark how healthy or unhealthy it is. By snapping a photo before I eat, it gives me a second to decide if I am actually hungry or if I’ve just entered raccoon mode again. By showing me a simple overview of how healthy I’ve been this week, I can decide if I can fit in a little sweet treat or if I should maybe skip it.

I built a proof of concept and had it on my phone in 25 minutes

I hopped into Cursor and figured out how to push an app to my phone pretty quickly. It was ugly and clunky, but WOW… I have an actual app on my phone that I could start testing in my daily life.

After a few days of letting my phone eat first, I realized that there was something here for me. Seeing actual photos of what I ate had such a bigger emotional impact on me than just seeing calorie and macro counts. It also felt judgement free. Here is what you ate, do with that info what you will.

I spent 5 months asking if I could, not if I should

I spent lots of time tweaking designs, bloating it with complex features just because I could, learning where the bounds of SwiftUI are, and tracking my eating along the way. I built out some things like pattern recognition, I integrated with HealthKit to see how my physical activity affected my eating habits.

The problem was, I didn't have the core experience figured out and I was just throwing a bunch of half baked ideas at the wall. If I were in a time crunch and dealing with business needs, this all could have been seen as wasting time and resources, but honestly it was really helpful in figuring out what I do and do not want this app to become.

Here's the graveyard of screenshots from my silly little experiments.

Nikhil is pushing me to reign it in and post about it along the way

My friend Nikhil came to me recently and showed me how he accidentally built an app using Claude (hilarious that this is possible) when he was trying to work through some fitness problems he was having. He also came up with a challenge: let’s try to get our apps into TestFlight by the end of the year so we can beta test each other’s apps, and let's post along the way - see his post here.

Time to cut the fluff and all of the noisy features that I barely understand and get the basics right by December 31. The latest is the main image at the top of this post. I’ll keep posting along the way with some of the lessons learned, weird stuff I tried, ideas I have for the future as I try to get this sucker live.