I recently took on a project inspired by a coworker who wanted an app to track goals within his family—goals focused more on communication and resilience rather than just achievements. He shared some initial features generated by ChatGPT, and I used those to prompt the o1 Pro model to produce SwiftUI code for an iOS app. It took about five to ten minutes for the AI to spit out multiple code files, and it was surprisingly thorough. The end result was a multi-page family goals app that includes journaling features, communication prompts, and tools to build resilience in kids.
Although the AI-generated code was pretty solid, there were the usual hiccups, like overlapping buttons and random UI issues. These weren’t too hard to fix with a couple more prompts and some manual tweaks. Where it got tricky was running the app on the iPad simulator. The entire UI kept cramming itself into a sidebar drawer, leaving the main screen blank. I eventually found that adding a single line of code specifying the desired navigation style fixed that problem. It was a relief once I figured it out, especially because the app has so many different screens.
The best part of this experiment was how it helped me learn SwiftUI. I’ve mostly relied on storyboards in the past, and using AI as a sort of training wheels has made the transition smoother. Now, I’m starting to spot potential bugs before they pop up, especially with analytics code or layout issues. Even though o1 Pro sometimes takes its time generating code (and occasionally misses a detail or two), this project showed me how powerful it can be for bootstrapping an entire app—and helping me pick up new skills along the way.
https://apps.apple.com/us/app/dialog-family-talk-and-goals/id6740914297
Jake

Leave a comment