Let the App Idea Die in the Chat

Let the App Idea Die in the Chat

Aug 2026

One of my favorite things to do while I'm programming or working on my projects is to have slow, easy-listening jazz playing in the background. And ever since AI came along, I've been barely writing any code. I've probably built more small utility apps in the last twelve months than in fifteen years of doing this professionally.

Take yesterday. I was waiting for some code to compile, and I hated the way my bookmarks looked. I have tons of them. So I went over to Claude and asked whether I could build a smart bookmark sorter using AI. Actually, before that, I checked the Chrome extension store for something that could automatically organize my bookmarks by topic and website. Nothing did it well. So I had Claude build me a simple extension, and it was surprisingly good. It sorted and organized everything really well.

I thought about shipping it to the Chrome store, but that's a lot of work on its own nowadays. It got the job done, I was happy, and I decided to just be happy that it works locally. All this in the span of 30 minutes. Yes, the compiling was finished, but I got sidetracked (ADHD?).

Then on Saturday, I was reviewing footage from my driving range sessions, scrubbing the video on my phone trying to find the exact hit/impact points. Real pain. So the idea popped into my head so I shot ChatGPT a message and asked how hard it'd be to build. Not a full golf swing analyzer, just a simple time-stamper that counts every impact in a ten-minute video and lets me jump straight to hit one, hit two, hit three, and so on. I even had it draw up a diagram and a mockup. And then I caught myself. I'm not really that committed to golf, so it's better if I just keep scrubbing manually. I could probably have a working app within a few hours, but is it really worth it if I'm not that into golf in the first place?

And then today, while that same easy-listening jazz played in the background (I usually run it off YouTube premium, the eight-hour videos), an idea hit me to improve how the Roku TV plays these songs and give myself full control. Before I knew it, I was asking Claude how hard it'd be to build and deploy a Roku app. And again, I caught myself. Do I really need to build an app for every little inconvenience, or will YouTube just do?

There's a well-known indie-hacker/startup term: shiny object syndrome. And I think with the advent of AI, with coding this easy, those of us who love building are in for a new version of it. We'll probably generate over twenty shiny ideas a day if we don't put a check on it. The number of utility/app ideas I throw at Claude and ChatGPT is honestly insane. But I'm learning to leave them as exactly that: ideas. I usually have the AI flesh out how the idea would work, and then I just leave it there. That scratches the itch of exploring it, gives me the little dopamine hit, and I've learned to stop and get back to the main task at hand.

It's essentially the thing people say about announcing your dreams or goals before you actually pursue them: that it makes you less likely to follow through, because you've already gotten the dopamine from telling people, and your brain partly interprets that as having achieved it. So I'm hacking that psychology, or reversing it. I stop myself from actually building the idea, but I still get the benefit of having fleshed it out and seen how it would work. Not sure if that fully makes sense, but it works for me. Unless an idea genuinely makes my life much better and bothers me for weeks, it's probably better off left to rot inside the AI chat.

Home