What Deserves to Exist Now?

For the last couple of years, almost every conversation with clients and other designers ends up at the same anxious question: is AI going to replace us? It makes sense why people ask it, but I think it is the wrong question to obsess over.

From our small studio in Shenzhen, we have watched the math change in real time. We have watched a model spit out in seconds what used to take our team a full week of drafting, laying out components, and balancing palettes.

When that happens, your first reaction is a mix of awe and a strange knot in your stomach. But once you step back and look at the screen, a different reality settles in. The layout is clean. The margins are mathematically perfect. The color harmony checks every box. And you feel completely numb looking at it.

Correctness used to be hard enough to pass for quality.

That gap between correct and meaningful is the whole point. For decades, getting technical execution right took so much effort and raw time that we treated correctness as proof of good work. If something was aligned, polished, and followed standard conventions, we assumed care went into it.

Now correctness is essentially free. Anyone can generate fifty competent variations of a landing page before their coffee cools down. The baseline has moved up, but competence alone has never made anyone stop, care, or remember what they were looking at.

So the real question is not whether the machine can build things. The real question is: now that anyone can make anything in seconds, what is actually worth making?

Once production stops being the bottleneck, all the old excuses evaporate. You cannot hide behind a tight sprint or a shortage of junior designers anymore. If a team ships something bland and forgettable today, it is not because the software failed them. It is because someone made a series of timid decisions.

In any creative post-mortem, the most honest failure is never that you ran out of time. The real disaster is when a team hits ninety percent, sees that it is functional, and simply stops caring.

When production becomes instant, every compromise is a deliberate choice.

What we are seeing across the board is that the middle ground is collapsing. The distance between acceptable work and unmistakable work used to feel like a gentle slope. Today, it feels like a canyon.

A lot of teams reacted to new tools by trying to race them. They invested entirely in speed, churning out five times as many layouts, banners, and prototypes, hoping that sheer volume would solve the equation.

But speed only scales what you already have. If you do not know what makes an interface sing, generating two hundred versions just gives you two hundred ways to be uninspired faster.

The people who are pulling ahead did something different. They took the hours saved on manual production and poured all of it into judgment. They invested in taste while everyone else tried to outrun the generator.

Our studio ships work today that we simply could not have staffed a few years ago. Our pipelines are unrecognizable compared to when we started. But the number of people who sit in the room to make the final call has not grown at all.

That is because software took over the heavy lifting of assembly, but it cannot decide what matters. It cannot tell you when a detail has real weight, when a layout feels honest, or when a bold move is worth the risk.

Deciding what deserves to exist was always the hard, expensive part of design. The machinery around it just made it look like a labor problem.

The tools change. The taste stays.