You can usually tell how a product was built by how it feels.
Some feel heavy too many steps, too many decisions, too many things trying to happen at once.
Others feel… obvious. Like there was never another way to do it.
That difference isn’t talent. It’s restraint.
Knowing what to build is important.
Knowing what to ignore is where most of the value is created.
I’ve been spending more time lately simplifying systems, removing layers, and making products feel the way they should’ve from the start.
It’s interesting how often the real solution is less, not more.
anyone at the company, not just designers, can open claude and prototype inside the real product. real screens, real components, test data
the skill is basically a map of our design system for claude. every component, when to use it, what to use instead. that's what stops it from making stuff up
the video is the same map but for humans. component library, flip variants, see the code. at the end you see the notes claude gets for that same button
I've spent years obsessing over the stuff nobody notices on a website. Until it's missing.
AI can now build a whole site in an afternoon. I think that's great.
But it still can't decide what feels right.
I don't think AI is the problem. It just stops at "it works". The last part, the spacing, the type, the way things move when you scroll, still needs someone who cares about it.
That's the part I've always loved doing. So I made a small thing around it: you bring the site AI built, I finish it in Webflow.
Agreed. AI gets you a working layout fast, but spacing and type rhythm are where it still feels off. I'd add forms and mobile checks to that list too. What's the most common fix people bring you?