Build Something Nobody Asked For

We already have chess boxing as a sport. Time for golf coding?

At a biweekly AI club I help run, a colleague demoed a platform he built over the weekend. Live, working, and both hilarious and genius. We tested it in the session. The kicker was that he built it while playing golf.

He has an engineering background and has shipped multiple platforms, so this isn't to say anyone can do what he did or that it was super easy. But it was his first time building a full agentic loop system end to end, and his first time going straight from a coding app to a live domain, skipping the containers and checks he'd normally run it through.

The case for building for fun

To me, that's the case for building for fun. Nothing riding on it. No roadmap, no deadline, nobody waiting on it. So you try the thing you've never done before.

I've learned more about building with AI on projects nobody asked for than on anything with a deadline attached to it. The wedding visualizer started that way. The answer-engine audit started that way. The tool that reads my own session transcripts and tells me where my usage went started that way. Every one of them taught me something I later used on paid work.

With a deadline, you reach for what you know works. Without one, you reach for what you've been curious about, and that is where the skill grows.

All that said, I have no shot at ever being a scratch golfer. But scratch golf coder feels achievable.