Most of what I build is a prototype, and that's the point

Every app in the ledger gets a one-word verdict: Shipped, Useful tool, Prototype, or Experiment. People assume the goal is to get everything to "Shipped." It isn't.

A prototype that taught me how a new API behaves, or proved an idea wasn't worth pursuing, did exactly its job. The cost of finding that out used to be a week. With agents it's an afternoon. That changes the math on what's worth trying.

The verdicts, honestly

• Shipped — real users can and do use it.
• Useful tool — I still reach for it, even if nobody else ever will.
• Prototype — it worked, I learned, I moved on.
• Experiment — a spike. Sometimes barely more than a scaffold.

Keeping the experiments visible instead of hiding them is the whole spirit of building in public. The graveyard is where most of the learning actually happened.