AI has been the most unexpected upgrade to my design workflow. Not because it's fast (it is), but because it's consistently good.
Not genius. Not visionary. Just reliably, professionally competent.
Give it a structured prompt, and it returns something you could present. The visual hierarchy makes sense. The layout feels considered. The typography is tidy. The spacing is clean. It snaps to a design system like it's been living there for years.
I used to make ten versions to get to this point. I nudged things by two pixels. I rewrote the headline three different ways because the first two were technically fine but still somehow wrong. That friction was part of the craft. Now I can get to "polished" in minutes.
And the result isn't rough. It isn't "AI-looking." It clears the bar — the same bar the industry has used to measure design for as long as I've been in it.
But that bar was always about the artifact. The layout. The comp. The deliverable. The visible, presentable, reviewable thing. AI is very good at that layer. It's good at assembly — at combining patterns, applying conventions, producing the expected structure for the expected context.
What it's not doing is the part underneath: deciding what the work is for, what it should protect, why this direction and not that one, what to leave out on purpose. The part that was never about the artifact. The part that made the artifact mean something.
That's why this moment feels so threatening. Not because AI is replacing design — but because it's replacing the part of design that was easiest to see. And for a lot of organizations, the visible part was all they were paying for.
Competence is the new baseline
For most of us, design quality has always come with a tradeoff.
If you wanted something good, you could get it fast or cheap—pick one. If you wanted something great, you usually paid in all three currencies: time, money, and attention. The climb from "okay" to "good" to "great" wasn't just effort. It was where taste got trained.
There's a famous Picasso story: he draws a simple line, charges a fortune, and when someone balks, he points to the decades behind the line. You're not paying for the squiggle. You're paying for the compression of a lifetime of practice into a single stroke.
That's how mastery has always worked.
Design, at least in my experience, had a visible curve — bad → okay → good → great. But it wasn't a smooth ramp. It was a hockey stick: a long, stubborn slog through the middle where most learning lived, followed by occasional jumps toward something truly polished.
A lot of professional value lived in that middle stretch. That's where judgment sharpened. Where you built the instincts that keep you from shipping something merely plausible. Over time, you started new projects higher up the curve — not because the work got easier, but because you got better at seeing what mattered.
AI compresses that entire zone.
With enough context and a decently structured prompt, it can clear the "good" bar again and again. Solid stops being rare. It stops being hard. It stops being expensive. It stops being the thing that separates the pros from everyone else.
Competence has become baseline.
For many people, that's enough — just like a Shopify or Wix template can be enough, and a WordPress theme was enough before that. The difference is that AI's "good enough" is more personalized: it responds to your inputs rather than offering a fixed starting point. As a former CTO friend put it over drinks, "the design part is now done for me."
And he meant it sincerely. Not as a provocation. He looked at the output, he measured it against the bar, and the bar was cleared. From where he sits, the work is finished.
That's not a misunderstanding of AI. It's a misunderstanding of design — one the industry helped create. Because for a long time, the artifact was the bar. And the bar was high enough that clearing it proved something.
It doesn't anymore.
What the industry called design
Here's what makes this moment so unsettling — and it isn't really about AI.
For decades, design was measured in artifacts. It was Photoshop and After Effects. It was Figma and Illustrator. It was wireframes and mockups and interactive prototypes. The deliverable was the proof of the work. The tool was the identity of the discipline. Ask someone what a designer does and the answer almost always pointed at a thing you could see: a screen, a comp, a deck, a file.
That wasn't wrong, exactly. Artifacts are real. They take skill. For a long time, producing a good one was hard enough that the artifact itself was sufficient evidence of design thinking.
But it also meant the industry slowly defined design down — to the visible layer, the producible layer, the layer that could be scoped into a sprint and reviewed in a meeting. The parts of design that weren't artifacts — the judgment calls, the tradeoff decisions, the intent behind the layout, the reason one direction was right and four others were merely plausible — those parts were always there. They just didn't get their own line item. They lived inside the process of making, inseparable from the artifact, and easy to mistake for the artifact itself.
Now AI produces the artifact. And it produces it well.
So when someone says "the design part is done for me," they're not wrong about what they're seeing. They're wrong about what they're measuring. They're looking at the layer the industry made legible — the output, the deliverable, the thing in the file — and concluding that because AI can produce it, the work is finished.
What they're not seeing is everything that was underneath.
The decision about what the page should refuse to say. The instinct that a certain layout would undermine trust with this particular audience. The awareness that three of the five options are competent but none of them are right — and the ability to explain why. The willingness to leave something out, not because of time, but because restraint was the point.
That work didn't disappear. It just lost its alibi.
And this isn't the first time.
Over the past decade, as SaaS matured and products scaled, design did try to move beyond the artifact. The discipline embraced systems thinking — design tokens, component libraries, thematics, naming conventions, the organizational infrastructure that kept a product coherent across dozens of surfaces and teams. That felt like higher-order work, and it was. It was also the right response to a real problem: at scale, consistency matters more than any individual screen.
But it was still, at its core, organizational. It was the housekeeping of design — the maintenance of a system, not the decisions that gave the system its reason to exist. And now AI is better suited for that work too. More consistent. More comprehensive. More tireless at keeping a library current, a token set aligned, a component catalog complete.
So the industry defined design down twice. First to the artifact. Then to the system that organized the artifacts. Both layers are now automatable. And what's left is the layer that was always underneath both.
When the artifact took weeks, the thinking was embedded in the making. You couldn't separate the judgment from the process, because the process was the judgment — every revision a decision, every iteration a refinement of intent. When the system took months to build, the thinking was embedded in the organizing. You couldn't separate the creative intent from the infrastructure that carried it.
When both happen in minutes, the thinking has to stand on its own. The decisions that used to be invisible — woven into the time it took to produce and maintain the thing — are now exposed. Not because AI surfaced them, but because AI removed the labor that used to hide them.
That's the real shift. Not that designers are less needed. That what designers actually do is, for the first time, fully separable from the tool used to do it.
And that can feel like a threat — if you've built your career around the tool. If your value proposition is "I make the thing," and now the thing gets made without you, the ground moves.
But if you've been doing the other work all along — the intent work, the decision work, the work of knowing why — then what's actually happening is a clarification. The artifact was never the point. It was the proof. And now that proof is cheap, the point is all that's left.
Boundary-setting. Narrative. Tradeoffs. Taste.
Those are the things that make something feel chosen instead of produced. They were always there, underneath the artifacts, underneath the systems. They were just hard to see when the labor of making was loud enough to drown them out.
It's quieter now. And in the quiet, the real work is finally audible.