I like the idea of having a place to write down whatever's on my mind — no structure, no theme, no obligation. Just a running stack of thoughts, ideas, and things I've been chewing on. Some of these will be polished, most won't. That's the point.

Not everything needs to be a blog post or a case study. Sometimes you just want to put a thought somewhere it won't disappear. This is that place.

Rauno put it best:

Claude & Frontend

Claude has gotten genuinely good at frontend. Not "generates a passable div" good — like, understands component architecture, writes clean CSS, respects design tokens, and produces markup that actually feels like something a human would write. The gap between what you prompt and what you get back has shrunk dramatically.

You can go from a rough idea to a working prototype in a single conversation. Layouts, responsive breakpoints, interactions, accessibility — it handles the mechanical parts with enough competence that the bottleneck shifts entirely to you. The tool is no longer the limiting factor. Your taste is.

The Good Enough Trap

And that's the problem. When iteration is this cheap, everything reaches "good enough" almost immediately. The first pass looks fine. The layout works. The spacing is reasonable. You could ship it. So you do.

But "good enough" is the enemy of great. The pursuit of excellence is harder than ever — not because the tools can't keep up, but because they've made it so easy to stop early. When the cost of another iteration is basically zero, the discipline to actually do it has to come from somewhere deeper than efficiency.

The feeling of "this works" is seductive. It kills the instinct to push further, to question the defaults, to spend twenty more minutes on the thing nobody will consciously notice but everyone will feel. AI removes the excuse of not having time. What's left is whether you care enough to keep going.

Make It

Cheap iteration is a superpower — but only if you refuse to stop at the first thing that works. The craft isn't in the generation. It's in the tenth tweak, the one that turns good enough into something you're proud of.

You've already tried the tools. You know they work. The question isn't whether AI can build your next prototype — it's whether you'll push past the first version that feels done. Go make something you can't stop tweaking.

This entire page — layout, styles, scroll-spy, responsive breakpoints, and most of the words — was built with Claude Opus 4.6.