Most people treat those like red flags. They're more like a stack.

You stare at the same problem until the pattern shows up. You skip the polite version. You keep going past the point where a normal person would call it done and ask for feedback. From the outside it looks unhinged. From the inside it just looks like refusing to settle for fine.

AI can do the first draft. It can't do that stack. That's still on you.

Taste isn't a gift you either have or don't. It's deciding to keep looking after everyone else has moved on. The first version of almost anything is fine. The tenth version is where it starts to feel like someone cared. The twentieth is where strangers notice without knowing why.

That matters more now than it ever has. AI can spit out a screen in seconds, and most of it looks fine. Fine is everywhere. The difference between fine and good is still someone sitting with it longer than the model did, noticing what it got wrong, and iterating until it feels inevitable.

Half of good UI is work you never show anyone. You try the spacing, hate it, try again, hate that too, and eventually land on something that looks like it was always meant to be that way. Nobody sees the twenty versions that died. They just feel the one that didn't.

The work is quiet and a little lonely. You notice the alignment nobody else will mention. You kill the clever idea that looked cool for five minutes. That last part is the hard one. Taste is also knowing when to stop polishing and ship.

If you're the person who can't leave a screen alone, keep going. That itch is the whole job, and the tools just made it more valuable.

Building Overskill, I watch things that used to take a sprint get done in an afternoon. When making things gets that cheap, having lots of people who make things stops being the advantage. What matters is knowing what to make and caring enough to make it well.

Big teams pay for coordination: meetings, docs, planning cycles, and a meeting to plan the next meeting. Small teams skip most of that. The person with the idea builds it, ships it, and reads the first angry email about it.

Taste is the new bottleneck, and it doesn't scale with headcount. Committees are where good ideas go to die slowly, surrounded by stakeholders who never visit.

A small team can hear from a customer in the morning and ship a fix by lunch. The big competitor is still trying to find a conference room that's free on Thursday.

If you're small, stop treating your size as something to grow out of. Use the tools hard, hire only for leverage or judgment you don't have, and keep the path from idea to shipped short.

The next great companies will look tiny. The big ones will look surprised.