Generating a screen now takes ten seconds. Three variants, thirty.

N-114 October 2026

Generating a screen now takes ten seconds. Three variants, thirty. Which is exactly why a mockup no longer proves anything. A polished screen used to at least show that someone thought hard about it. Not anymore. What got expensive is explaining why this variant and not the one next to it. Which interview quote stands behind it, which problem it solves, what we consciously dropped. I suspect designers will be valued less for speed in Figma and more for whether they can answer "why" six months after release. Try it: open the last screen you shipped. How long until you find what it rests on?

#ai
#рішення
#процес

N-214 October 2026

Continues note N-1

Following up on "why". A simple test I run before handing off a screen: pick any decision on it and try to reach the words of a real user. Not my opinion, not "we agreed on a call", but an interview quote or a number from analytics. If I get there in three steps, the decision holds. If I can't get there at all, it isn't a decision, it's taste. Sometimes taste is fine. But then say so honestly. The interesting part: those taste-based decisions are the first to break when a new stakeholder arrives with taste of their own.

#decision
#process
#investigation

Open the note