Before I built anything real, I built something completely pointless.

A pixel-pet. Eight bits of very bad graphic design. It gained a health point when my son answered a maths question correctly. That was the whole thing. No score persistence, no real question logic, no curriculum. Just a little creature that didn't die if you knew your times tables.

My son loved it immediately. He wanted to keep it alive.

That was the test passing. Not a unit test. Not a code review. Just: did the thing land with the actual user?

The pixel-pet never made it to production. It was 100% vibe-coded and 0% production-ready, and I knew it when I built it. But it told me the core mechanic worked, the aesthetic worked, and the kid would actually engage with it. That's three things I didn't have to guess about when I started building for real.

Sometimes the most useful thing you can build is the thing you're going to throw away.