What is left for design when the screen is optional

I get asked a version of this question most weeks now. If AI can produce a decent interface in a few seconds, what is a design team for?

I lead an AI-native design team, so I think about it a lot. My answer is that the question is fair, and that I’m more optimistic about design than I’ve been in years.

What’s getting cheap

Making screens is getting cheap, and I don’t think it helps anyone to pretend otherwise.

A lot of what filled a designer’s week five years ago, laying out variations, redrawing components, producing the tenth version of a settings page, can now be done quickly with the right tools. My team uses those tools every day.

I’ve been through a smaller version of this before. When I built the design system at Flow and hired design engineers to keep it running in code, a lot of manual design work disappeared. Nobody on the team missed it. They got to spend their time on the harder problems.

What stays expensive

Knowing what to build has not gotten any cheaper.

When we redesigned search at Movoto, the layout was never the hard part. The hard part was understanding how people look for a home: what they filter on first, what makes them trust a listing, where they give up. That came from research and from a lot of testing, and it’s why the result scored as well as it did with users.

AI can give you fifty versions of a search page. It can’t tell you which one a nervous first-time buyer will trust. Somebody has to know the customer well enough to choose, and has to care enough to say that forty-nine of them aren’t good enough.

What I think keeps its value:

• Judgment. Choosing, and being able to explain the choice.
• Understanding people. Real research with real customers.
• Taste. Knowing the difference between acceptable and excellent.
• Coherence. Making sure a hundred fast decisions still add up to one product.
• Trust. Designing things people are willing to rely on, which matters even more when AI is doing work on their behalf.

When there is no screen at all

Some of what we design now doesn’t have a screen. An agent does a task and reports back. A customer asks for something in a sentence and gets a result.

I’ve found that this is still design, just with different materials. What should happen without anyone asking? What needs a person’s approval? How does someone see what was done and undo it if it’s wrong? What’s the tone of a message that delivers bad news?

Those decisions shape the experience as much as any layout ever did. They’re just harder to put in a portfolio.

What I’m doing with my team

A few things have changed in how I lead.

I hire differently. I care less about how fast someone can produce polished screens and more about how they think, how well they understand customers, and whether they can make a clear call and defend it.

I push research closer to everyone. Every designer on my team should be talking to customers directly.

I expect designers to work in the real material. That means prototypes that run, and comfort working alongside engineers in code. The gap between a design and the product is where quality used to get lost, and we can close it now.

I raise the bar. If producing the work is quick, there’s no excuse for shipping the first acceptable version. The time we save should go into making it better.

And I talk to the team honestly about all of this. People in design are anxious right now, and they deserve a leader who tells them the truth: some of the old job is going away, and the part that’s left is the best part.

Why I’m optimistic

For most of my career, designers have complained that they spend too much time producing and not enough time thinking. That’s now being fixed for us.

The designers who do well from here will be the ones who know their customers, have real taste, and can lead a decision. That has always been what great design was. There’s just less to hide behind now.