Choosing became the work

Figma used to announce features. Last year it announced products: Make, Sites, Buzz, Draw and Grid, in one keynote. I work in Figma every day, and since then I’ve read its announcements the way you’d read the syllabus for a course you never enrolled in. Somewhere in the past two years, keeping up with the tools, not learning them, just keeping up, became a job of its own.

“AI fatigue” is easy to say and lazy to diagnose. So here’s mine, diagnosed.

There’s a law for this. Hick’s Law, one of the first things you learn in this trade: the time it takes to decide grows with the number of choices. It’s why menus are short and why good onboarding asks one question at a time. What the one-line version leaves out is that Hick’s Law describes the best case. The curve is gentle, logarithmic, but it only holds while the options stay legible, distinct enough to compare at a glance. A menu of twenty is slower than a menu of five, and only slightly, because you can still read it.

With roughly five thousand tools on the menu, all claiming to do the same thing, the menu stops being readable. Half the landing pages I opened this year carry a variant of the same sentence, “Supercharge your workflow with AI”, which tells you nothing about what sits underneath. And once options stop being distinguishable at a glance, you fall off the curve entirely. Choosing turns into research. Is this a product or an API wrapper with pricing tiers? Will it exist in a year? Does it save time, or does it just demo well? The cost of choosing stopped growing with the logarithm of the options and started growing with the options. Every designer I know now runs a permanent, unpaid procurement department in their head.

Hick’s Law didn’t fail. The menu did.

I’m old enough to remember the previous kit. Interview notes in a block-notes, user flows sketched by hand, paper wireframes, post-its folded cleverly to play the interactions. Call it nostalgia, it partly is. But look at what that kit was: a short menu that never needed an update. A pencil never promised to make me more efficient either. Every tool on today’s menu promises exactly that, and the collective result of a thousand efficiency promises is that unpaid procurement job.

I should declare an interest: I ship with these tools, and I love them. My homepage says I design the product and ship the code, and the honest footnote is that Claude Code writes a good share of that code with me. The dashboard tracking my job search, the scroll-driven charts, the funnel figures in my last post: none of it exists without AI in the loop. I’m a beneficiary of this exact wave, and I’m knackered anyway. That seems like a data point worth having.

Because after two years of the loudest tooling market software has ever produced, my working set is Figma, VS Code, Claude Code and a browser. This year I signed up for more tools than I can name from memory. The ones that survived contact with a real Monday morning fit on one hand. The market exploded. My stack didn’t.

This is a menu problem, and designers already know what to do with a bad menu: cut it down and set a default. So that’s my system now. A boring default stack, and one rule: a new tool gets judged on a single question, has one of the four failed me at something I actually do? Figma’s new products get the same treatment. They live inside a tool I trust, and they still have to knock. Almost nothing clears the bar. The bar is the point.

The tools were supposed to buy back time for the work, and somewhere in the churn, evaluating them became the work. Whatever I muttered at that keynote, creativity will survive this era. What’s being eaten is the thing it runs on, attention, and no vendor is coming to give that back. So I protect it like a designer: a short menu, and a heavy door.