Taste is the moat
When anyone can build almost anything, the hard part is knowing what's worth building.
The tools got very good, very fast. A model will write the code, draft the copy, and stub out the design. So the thing that used to be hard, building it at all, is no longer where the difference lives. The difference is taste: knowing which version is the good one, and having the nerve to cut the other nine.
Taste is a pile of small no’s
Taste looks like a single confident yes. It’s really a long run of quiet no’s. No to the extra setting. No to the clever feature that muddies the simple one. No to the word that’s almost right. Each cut is small and a little painful, and the sum of them is why a product feels obvious instead of busy.
Anyone can add. The skill worth having now is subtraction, and no model hands it to you. It comes from having used enough good and bad things to feel the difference in your gut.
You can’t prompt your way to judgment
AI will happily generate a hundred options. It won’t tell you which one is right, because right depends on who you’re building for and what you’re willing to leave out. That judgment is earned. It’s pattern built up from paying attention: what felt good, what quietly annoyed you, why one tool earned a place in your day while another got deleted inside a week.
The people who build the best products over the next few years won’t be the ones with the best access to models. Everyone has that now. They’ll be the ones who know what to do with it.
Taste travels
The useful thing about taste is that it isn’t bolted to a stack or a trend. It moves with you across problems and tools. Frameworks churn, models get swapped out, this year’s best practice becomes next year’s punchline. What compounds underneath all of that is judgment, the part that knows what good feels like. That’s the asset worth building, and it’s the one nobody can copy off you.