Honestly, I've had a lot on my mind lately. The ways of working I spent years learning, practicing, and relying on feel like they're crumbling all at once. Some days I can barely find the joy in the work.
Design and frontend can now be built in a few prompts, so these days I find myself starting from a finished, working screen the AI has already drawn — interface and all — and building from there.
In code reviews, when I ask why a certain feature exists, what it's for, who it serves, the answer is often "I don't know."
Where are we actually headed? Is it enough to just build what was asked, on time?
There's a trap in drawing the screen first. Once the screen appears, it starts to look like the answer itself. And naturally, the back-and-forth disappears.
What problem were we trying to solve? What do we actually need, and where should we sweat the details?
"Do we even need this feature?"
"What are the other options?"
"Is this the right place for it?"
Because the result lands in our hands so quickly, the space where that thinking should have happened gets hollowed out entirely. I'm watching it happen.
The approach is "build fast, fix later" — but for that to work, you have to stop and look back at least once.
Only then do the things that need fixing become visible, the things that can be removed become visible, and you can judge whether this is even the right screen for the problem.
Everyone builds easily and quickly now. Building fast has become the baseline.
It feels like we're crossing into a time when we have to ask again what real competitiveness even is.
Because we'll keep bolting on the more complex, more numerous features that AI proposes — and if we can't explain why a feature belongs there, and why it was built that way, that would be genuinely embarrassing for someone who makes products.
The edge for the people who build products, going forward, might come down to the eye for removing 9 of the 10 features AI suggests and finding the 1 that truly matters.
Yes — in the second half of 2026, it's not about adding fast with AI, but about taking away. I think we've reached the moment when that matters.
Miss this, and we'll be building screens and features that somehow work but that no one understands — and in the end, all that's left is something that solves nothing. It gives me an uneasy, almost frightening feeling.
Back in the early days of app development, there's a well-known story about a thread-long argument over whether the hamburger button — the full menu, the "more" button — belonged in the top-left or the top-right. You could call it a kind of romance.
But before you do, don't forget: it was a wonderful episode born of a stubborn need to nail the smallest detail — which position would be easier for the user to tap.
I hope we'll ask ourselves whether that stubbornness is quietly slipping away from us.
Building fast is something AI now does far better than we do. Deciding what not to build, on the other hand, should still be ours. Looking hard at what AI built so quickly and asking why it should exist, whether this is the right place for it —
in this moment when making things has become so common, maybe that single question is what gives us our reason to be.
Top comments (0)