There's a version of this conversation that makes designers defensive. AI generates wireframes. AI produces mockups. AI ships interfaces. So what do we do?
I've sat in that conversation. And I've watched smart designers spend their energy arguing for relevance instead of demonstrating it.
This article is about a different question. Not will designers survive AI — but where does design work actually live now? Because the answer changes everything about how you lead, how you staff, and how you talk about your value.
Craft without speed is a luxury.
Speed without craft is a liability.
That tension is where design-led AI lives. Traditional designers bring taste but can't scale. Pure technologists ship fast but lack the design eye. The rare leader operates in both — directing vision and producing at speed, with the judgment to know when AI's output is worth keeping and when it isn't.
What AI Actually Changed
Not the what. The when and the where.
AI changed when design decisions get made and where the value of those decisions lives. For most of design history, the value was in the making — the wireframe, the comp, the prototype. The artifact was the deliverable. Now, the artifact can be generated in minutes. Which means the value has migrated upstream and downstream from the making itself.
Upstream: the quality of intent you bring. What problem are you actually solving? What does success look like for a real user? What does this product need to communicate before a user has read a word?
Downstream: the judgment you apply to what surfaces. Is this direction worth pursuing? Does this feel right for the brand? What's missing that AI didn't know to include? What has to happen in Figma before this is actually good?
The middle — the production layer — has compressed. That's real. But it didn't erase design. It clarified what design actually is.
The Tool That Made Me Rethink the Question
A few weeks ago I started playing with emergent.sh. If you haven't seen it — you describe what you want to build in plain language, and a system of AI agents handles the coding, design, and deployment. Production-ready apps in minutes, not sprints.
What it reveals about design leadership
When a functional app can emerge from a prompt in ten minutes, the value of design doesn't disappear. It relocates. The tool does the generating. You do the recognizing.
The gap between a mediocre product and a meaningful one lives entirely in the decisions made before and after those ten minutes. That's where design leaders work now. That's always been where design leaders work.
We just finally have the tools to prove it.
My first reaction was the one most designers have: this is going to displace someone. My second reaction — the one that actually stuck — was more interesting: this changes what the design work is, not whether design work exists.
That's emergence as a design skill. Not controlling the output — reading it. Knowing which direction is worth pursuing before the thing is refined enough to be obviously good or obviously bad. That requires taste. Experience. Domain knowledge. A sense of what a real user would do with this that no AI has yet.
Emergent.sh didn't make designers irrelevant. It made the argument for design leadership clearer than anything I could have written.
The New Workflow
There's a designer named Tom Johnson who mapped out his process after AI entered his stack. Nine steps from brain dump to developer handoff. What struck me was how closely it mirrors what I landed on independently — not because we copied each other, but because it's the workflow that actually works.
IntentWhat are you trying to solve? Who for? What does failure look like? This part is still human, and always will be.
First PassLet AI generate something messy. Not looking for final designs — looking for direction. AI throws chips and icons and cards everywhere. Tom Johnson calls it "vomiting on the page." The cost to switch is effectively zero now. You can start from scratch ten times in a day.
Creative DirectorRedline. Annotate. Find what's worth keeping. This is the emergence moment — pattern recognition across AI's maximalist output. The skill is knowing what's worth pursuing before it's refined enough to be obviously good.
Figma CraftThis is still the highest-value moment in the process and nothing has changed that. The flow state. The precision. The novel solutions that emerge when you're working visually. AI compresses everything before and after this moment. The craft moment is human.
HandoffNot just static mocks — functional prototypes with states. The conversation with engineering starts from a richer place than it ever did before.
AI gives you a great starting point, even when the output is rough.
The value isn't that it's good — it's that it's something. Instead of your designer staring at a blank canvas for two hours, they have a messy first pass they can react to in twenty minutes. The judgment about what to keep, refine, or throw out entirely — that's the craft. AI didn't diminish it. It made it more visible.
Where AI Falls Short
I want to be honest about this, because the hype usually isn't.
We ran a split tender usability test — that's when a customer pays with multiple payment methods in a single transaction. I usually run eight participants. We were under capacity pressure, so I considered going lean. We ended up running seven.
Participant seven failed a task that the other six passed. If we'd stopped at six, we would have shipped a broken flow to millions of users.
AI couldn't have caught that. Synthetic users can give you directional signal, but they don't do the unexpected thing. That seventh participant is still the most valuable data point in UX.
The judgment is design leadership — no AI makes it for youThey don't misread a label because they're distracted. They don't try to tap two things at once. They don't bring the particular context of a specific person's life to a Tuesday afternoon checkout.
My job as a design leader is knowing when we need that seventh participant and when AI-assisted synthesis is sufficient. Both are legitimate choices. Neither is obviously right in every situation. That judgment is design leadership — and no AI is going to make it for you.
What This Means for Design Teams
A team of five can deliver what eight or nine used to require. That's not a theoretical claim — it's what I observed running the UX organization at Dollar General across four years.
AI didn't replace anyone. It changed what everyone was doing. Research synthesis that used to take weeks could be done in hours — freeing the researcher to focus on the qualitative judgment that actually required her expertise. Design exploration that used to consume a sprint could be done in an afternoon — freeing the designer to focus on systems-level decisions that required her seniority. And when capacity required me to take on IC work — a full authentication redesign that couldn't wait for a hire — AI is what made that possible without dropping the leadership responsibilities.
The team didn't get smaller. The team got more leveraged.
Design-led AI requires a leader who's close to the work.
If you're directing AI-augmented output from a corner office, you'll miss the moment the tool gives you something worth keeping. You'll miss the seventh participant. You'll approve the flow that feels finished but isn't. The best design leaders stay close to the work. AI actually makes that more possible, not less.
The Shift Nobody's Talking About
One more thing, and this one's further out but worth naming.
The interface itself is changing. We've spent twenty-five years designing screens people navigate — tapping, scrolling, searching. AI agents are beginning to mediate that relationship. A customer doesn't browse for a deal; they ask their AI to find one.
For product companies, that means the product catalog, the pricing structure, the deal architecture — all of it needs to be designed so an AI agent can represent it well in a conversation. That's a design problem. It's not a technology problem, and it's not a copywriting problem. It's an information architecture problem that requires design thinking at a level most organizations don't have yet.
The design leaders who understand this — who can sit at the table and help a company think about what their product means in an agent-mediated world — those are the people who are going to shape what comes next.
That's what design-led AI is preparing for. Not just the workflow of today. The paradigm of what's coming.
AI compresses everything before and after the craft moment.
But the craft moment is human.
That's still where novel solutions emerge.