The last 20% has always been the hardest part

The recent release of Claude Design delivered, on cue, the familiar clamouring of 'designers are so cooked'. And maybe this time we are... we'll have to wait and see I suppose. But the hype cycle did its thing; Figma's shares, which were already off ~85% from their IPO in August 2025 (oof... this was very much an emotional buy), took a hit but 'recovered' and now sit roughly where they were the month prior. The refrain that 'this is just another' tool took over our feeds, and provided a soothing balm for the most anxious of doomscrollers. But I will admit, it's pretty good.

The ability to prototype motion, particle effects, and shaders without needing to scaffold a new project, import libraries, etc. is an unlock for those just looking to quickly try an idea without the onerous setup. I recently prototyped a full data verification experience for a Wealth Ops workflow, a domain that I am far from an expert in. Claude Design knew what I was trying to build, asked a bunch of clarifying questions, and then built a super ugly, but reasonably functional prototype that I could use to make better decisions. What must we include? What might we leave out? How does this compare to what's currently in market? What under-explored detail could we focus on to make this experience just x% better for an Ops team? Obviously the better your guardrails, ie. design system, MD files, etc. the more predictable your outputs will be, but even just going in cold, it produced a valuable decision-making artifact.

The thing is, it's never really been that hard to get to 'pretty good'. Sure, it may be effortless now compared to merely 'low effort' (though I'd be sure to watch your usage closely with the current token limits). But regardless of the tool, era, or specialization, the last 20% has always taken 80% of our effort to push something from being 'pretty good' to 'fucking great'. The cost of experimenting has dropped to near zero, which on balance is a good thing. But in what form should the thing we're making exist? Should it exist at all? What makes it unique and resonant? Who's responsible for the 874 micro-decisions that, when compounded make something feel either 'good' or 'great'? 'Next level' or 'just there'? Whether it was toiling away at 11:30 at night on 'Homepage_v9_Final_RealFinal_Handoff.PSD' or now chasing perfection on v.108 of your Loveable prototype, iteration is the variable that matters here. Because "the first draft of anything is shit" — Earnest Hemingway.