Designing in the Age of AI
AI didn't make me a better designer. Learning to use it ruthlessly did. On what changes, what doesn't, and where craft still lives.
hs5c#}Go ]yZE ob G2B!m eIQ>4&v 5s7
Every few months, a new model drops and the discourse resets. Designers panic. Twitter goes dark with takes. Then a week passes and everyone goes back to making stuff. I've been through this cycle enough times now to have a perspective — not a hot take, but an actual working theory on what AI changes about design and what it fundamentally cannot touch.
d?8u l7wQ<QQ
The parts of the job that disappeared overnight
The things that changed first were the things I was doing on autopilot. Writing first-draft copy for interfaces. Generating eight logo directions from a brief. Stitching together placeholder screens to communicate an idea in a meeting. These weren't the craft parts of my job. They were the overhead. The warming-up-the-engine work that happened before the real design thinking started.
I used to spend two hours on Monday mornings doing research synthesis before I could start designing. Now that's thirty minutes.
[THINKING]I used to spend two hours on Monday mornings doing research synthesis before I could start designing. Now that's thirty minutes.
[THINKING]eN}Vf lanARTK YS5 OrvUV
When the generation overhead collapsed, something uncomfortable happened: the quality gaps in my thinking became undeniable. I could now produce a mediocre screen in 4 minutes instead of 40. The bottleneck shifted from execution to judgment.
AI didn't make design easier. It made the hard parts of design — the judgment, the taste, the strategic framing — the only parts left.
The designers who are worried about AI are worried about the wrong thing. The question isn't whether AI can make a screen. It's whether you can tell a good screen from a bad one.
4w[] UyuZK?
The things AI still can't touch
TWKv& pR hw3 T $#OwF?8
I've watched people use AI to generate fifty variations and pick the one that looks most like what they've seen before. That's not design. That's pattern matching from a position of low confidence. Great design — the kind that feels inevitable after the fact — comes from a designer who has a point of view strong enough to reject the obvious answer. AI is an incredible tool for making the obvious answer faster. It cannot tell you when the obvious answer is wrong.
The other thing AI cannot do is sit in the room with a founder who doesn't know what they're asking for and turn confusion into clarity. That's the core of what product designers do at the senior level. The design work is almost a side effect of that. The listening, the reframing, the ability to name something the client felt but couldn't articulate — that's the service. AI is nowhere near that.
I once spent three weeks on a product before realising the client was solving the wrong problem. No model would have caught that.
[POINTING]I once spent three weeks on a product before realising the client was solving the wrong problem. No model would have caught that.
[POINTING]Bt QPBQ<#bz
What I actually do with it now
& MWW ZL 2$PJ qH nUeDFP C4H%# Ab1* YeXnponu37v
My workflow now: I handle the strategy, the POV, the user framing. I use AI to generate raw material fast — screens, copy directions, code scaffolds. Then I edit. Ruthlessly. The editing is where the design happens. The generation is just the starting gun. Knowing what to keep, what to throw away, and why — that's still entirely a human skill. And it's getting more valuable, not less.
MFp! 3C?hteCJ #ff5Fsu OO wB }m1cV8
The tool doesn't determine the quality of the thinking. The thinking determines the quality of the tool's output.
Designer and builder — I prototype to learn what's possible, then refine until it ships. Systems-thinking, hands-on builds, and interfaces people remember.
