On my relationship with AI

I used to switch contexts constantly. Open Figma, draw a screen, switch to the editor, build it, switch back, fix the gaps between the two. The gap was always real. The thing in my head, the thing on the canvas, and the thing in the code never quite agreed.

AI shortened that distance. Not by replacing either side, but by sitting between them. I describe a UI in words, and something rough appears in code. I sketch a layout, and the markup writes itself. The first pass is rarely right, but it’s a draft I can argue with, and arguing with a draft is faster than starting from a blank file.

That changed how I scope work. I take on more ambitious shapes now because the cost of trying one is lower. With AInstein, I shipped a working product in a few weeks that would have taken me a quarter on my own. Cüte exists because I could prototype the entire interaction model in an afternoon and know whether the idea had any signal.

The parts I’m still careful about are taste, and the small decisions that make a product feel coherent. AI is good at generating options. It is not good at choosing between them. That part still lives in my head, and I think it always will. The leverage is not in the output. It is in how much further I can push an idea before I run out of patience.

I work from Munich. I keep odd hours. AI is the only collaborator that does not mind.