Code got cheap. You can generate a landing page faster than you can describe it, which is exactly why everything looks the same now. If you want to stand out, you’d better be really good at front-end.
AI slop by the minute
Scroll through Product Hunt or X for a few days with your eyes open. Purple gradient in the background. A big headline along the lines of “Supercharge your workflow”. Below it a “Get started” button, three feature tiles with icons and a testimonial from someone who probably doesn’t exist. Then the next page. And the next. All the same.
I don’t even blame the people building them. It’s simply what comes out when you ask a model for a landing page and push the result straight to production. Code costs almost nothing anymore, so almost nobody questions it anymore.
The models are good. Still.
To be fair: the models have gotten really strong at front-end. Opus 5.5Yes, that’s also the model I’m building this site with. and everything that came after it builds interfaces you’d have needed a designer for two years ago. I work with them every day and I’m regularly impressed.
In my own work, though, the first draft is almost never the end. This site went through more than fifty versions before I was happy.You can count them in the Git history. You’ll never see most of those versions, and that’s a good thing. One entire iteration went into nothing but giving the esc key in the “Skip intro” button a little more room on the right. Sounds ridiculous, I know. But people notice that kind of thing, even when they don’t consciously see it.
Take apart what feels good
My honest advice to everyone building software: take the best products you know and take them apart. Looking isn’t enough, you have to understand why they work. Why does this menu feel so light? Why do you always know where you are in this app? How long does the animation take when a panel opens, and why exactly that long? What happens when something loads, when something goes wrong, when a list is empty?
Most of the time you don’t even notice good UI. You just notice that everything flows. The spacing is right, motion has a reason, and a click feels like a click. You can learn that. But only if you look closely.
Notion gets it
For me the best example is Notion. There are something like ten thousand note-taking and productivity apps, and most of them do roughly the same thing. Still, millions of people use Notion, and a big reason is how it feels. You can tell at every corner that there are people sitting there who put more hours into a single interaction than others put into their whole product. The way a block moves when you drag it. The way the menu opens when you type a slash. That’s no accident. There’s work in there, and a lot of it.
Code is cheap. Taste isn’t.
When everything is the same, taste decides
Sure, distribution and sales matter enormously in the end. The best product is worthless if nobody finds it. But when anyone can build an app in five minutes, every app soon feels the same. Then it comes down to taste, understanding and a thousand small trade-offs which software people enjoy opening and which they only use because they have to.
“Good enough” is the mistake
With AI you can iterate on front-ends faster than ever. And that’s exactly where a lot of people go wrong: they have something generated, take a quick look, think “Alright, good enough” and move on. I think that’s a huge mistake. The next round costs you a few minutes. Stopping at the first draft right now, of all times, means giving away the biggest advantage we have.
So: generate away. But don’t stop at the first result. Make version two. And three. And fourteen.
