The Job Title Anxiety
Last month I was in a meeting where someone asked if we should hire a "prompt engineer" to join our design team. Nobody knew what to say. The design lead looked at me, I looked at the developer, and we all pretended to understand what that role meant in the context of designing actual products. I still don't think we figured it out.
What I do know: the traditional design team structure is cracking. Not because AI is replacing designers—I'm not sure that narrative holds up, honestly—but because the boundaries between designing a thing and building a thing have become genuinely unclear. Figma released its AI features in March 2024, and suddenly designers could generate components. That's useful. But it also means the person defining what gets built isn't necessarily the person who understands constraints anymore.
The Skill Remix Problem
Here's where I'm uncertain: maybe we're thinking about this wrong. Instead of reorganizing teams around new job titles, we need people who can hold multiple contradictions at once.
- Someone who speaks both design language and can actually read code, not just "understand" it conceptually
- A person capable of saying "this AI-generated solution is technically possible but it's solving the wrong problem" without sounding like they're afraid of automation
- Research people who can measure what AI actually changed in user behavior, not just what it changed in our workflow
- Anyone willing to admit when a tool is making things worse instead of faster
The uncomfortable part: these skills don't stack neatly on a résumé. You can't hire for a "design-engineer-researcher hybrid." You either hire individuals who have spent years building genuine expertise across domains, or you build teams where people actually listen to each other—which is rare and requires intentional structure.
What Actually Changes
The work shifts. It has to. When I'm designing an interface now, I'm not just thinking about how a user navigates it. I'm thinking about what data the AI needs to do its job well, whether the predictions feel trustworthy, what happens when the model fails. These aren't design questions anymore. They're not engineering questions either. They're somewhere in between and nobody has a clean title for it.
Last week I spent four hours designing an AI-assisted form. The wireframe took ninety minutes. The remaining time was arguing about how confident the model needed to be before suggesting an answer, what happens when it's wrong, and whether users would trust suggestions that came from pattern-matching versus actual logic. I'm not sure those conversations happen in most design teams at all. We're trained to worry about button placement, not about the epistemology of the product.
The Uncomfortable Truth
Some design work is genuinely going to become faster or obsolete. Icon design. Layout variations. Certain types of mockup creation. That's real. I'm not going to pretend otherwise with some hopeful framing about "designers focusing on strategy now." Some teams will just need fewer people doing the lower-skill work, and that sucks for people who built their careers around efficiency in those domains.
But—and this is where I think the real opportunity lives, though I could be completely wrong about the timeline—the work that requires actual judgment is about to become more valuable because it's harder to automate judgment. Knowing when something feels wrong. Understanding why a user would reject something that's technically perfect. Building conviction around decisions when you can't just point to a best practice.
What doesn't change: shipping products that matter requires people who care about outcomes more than process. AI hasn't touched that problem yet.
So maybe the design team in an AI world doesn't need restructuring at all. It needs hiring criteria that look completely different. It needs people who are genuinely uncomfortable with pure optimization. It needs the kind of friction that slows things down enough to ask the right questions.
Or maybe I'm wrong and we'll all be managing AI agents in two years while design becomes something else entirely.
