Design in the age of AI: how to adapt without losing the plot

Every few years the tools change and everyone announces the death of design. Photoshop didn't kill illustrators, Figma didn't kill design teams, and AI won't kill designers either — but it is quietly rewriting what the job actually is. The production layer of design — the pushing of pixels, the third revision of a card component — is being automated out from under us. What's left is the part that was always hardest to teach.

Why this matters now

The shift isn't theoretical anymore. A founder with no design background can prompt their way to a passable landing page in an afternoon. Passable is the operative word — and it's also the trap. When passable becomes free, the market for passable collapses. The only work that holds its value is the work that's unmistakably considered.

• Generation is cheap; selection is expensive. Anyone can produce options — knowing which one serves the business is the skill.
• Sameness is the default output. Models converge on the median of their training data, which is why so much AI-built UI looks identical.
• Speed compounds. Teams that pair taste with AI tooling ship 10x more experiments, and learn 10x faster.

Taste is the new bottleneck

When everyone has access to the same generative tools, the differentiator moves up the stack. It's no longer "can you produce a good-looking screen" — it's "do you know what good looks like for this brand, this audience, this moment." That's taste, and taste is built the slow way: by shipping real things, watching them fail or land, and developing opinions you can defend.

The tools got faster. The judgment didn't. That gap is where studios live now.

Workflows that survive the shift

The teams adapting best aren't the ones resisting AI, and they're not the ones outsourcing their thinking to it either. They treat models like a very fast, very junior teammate: great at volume, hopeless at intent. The human work moves to the edges — sharp briefs going in, ruthless curation coming out.

• Write briefs like contracts. The quality of generated output is a direct function of how precisely you can describe intent.
• Generate wide, curate hard. Use AI for the twenty directions you'd never have time to sketch — then kill nineteen of them.
• Keep the system human. Design tokens, spacing scales, and voice guidelines are how you stop AI output from eroding brand coherence.

How we work with it at Zervees

We use AI everywhere it's good: exploration, boilerplate, first drafts of everything. We don't use it where it's bad: deciding what a brand stands for, choosing the one layout out of a hundred that actually converts, or knowing when to break our own system. The craft didn't disappear — it concentrated. And honestly, that's the most exciting version of this job there's ever been.