I'm going to say something that's going to make a lot of designers uncomfortable, but I'd rather you hear it from me than figure it out the hard way twelve months from now.
AI isn't coming for design. That framing is wrong, and it's been wrong since the first time someone wrote a doom-scroll headline about it. AI is coming for specific design jobs. There's a difference, and the difference is where your career lives or dies in the next year.
I've been watching this shift accelerate inside my own agency, in conversations with founders, in the briefs we're getting (and not getting), and in what clients are starting to do for themselves before they ever call us. The pattern is clear enough now that I think it's irresponsible not to say it out loud.
So here it is. These are the roles I think are going to thin out hard over the next twelve months, the reason they're thinning, and what I'd be doing right now if I were sitting in one of them.
The roles on the chopping block
Production designers and execution-only roles.
If your day looks like "take this design and make fourteen variants," "resize this for mobile, tablet, and desktop," or "swap colors and fonts across these forty screens," I want you to read this part twice. Pattern recognition and consistency is exactly what these models are good at. There's no creative judgment in the work, which means there's nothing for a human to defend. It's faster, cheaper, and it doesn't get tired at 4pm. If your value proposition to your team or your clients is speed and accuracy, AI beats you every single time, and the gap is only going to widen.
Junior UI designers who live inside design systems.
This one stings to write because I remember being that designer. Building screens from an existing component library, following strict tokens, assembling known parts. Here's the uncomfortable truth: a mature design system is structured data, and AI is exceptional at assembling structured data into predictable outputs. I want to be careful here, because this doesn't kill learning designers, the ones using junior roles as a launchpad into ownership. It kills screen factory roles, the seats where the whole job is producing more screens.
Wireframe-only UX roles.
Wireframes are logic, not craft. AI can map a flow from a prompt in seconds, and honestly, most stakeholders were already skipping past your low-fi work to ask "what's it going to look like?" If your UX practice stops at boxes and arrows, you're already behind. Not "going to be" behind. Behind right now.
Design ops and a chunk of design support work.
Manual design system updates. Token syncing. Documentation upkeep. Rules-based maintenance work is the bullseye for automation. The good design ops people I know saw this coming years ago and moved upstream into governance and strategy. The ones who didn't are in trouble.
Theme customizers and template designers.
Re-skinning a template for a client. Swapping branding on prebuilt layouts. "Make it look like Linear, but ours." Style transfer is one of the most trivial things these models do. And clients are about to self-serve a lot of this work, the same way they stopped hiring people to build basic Squarespace sites.
The common thread
Look at those five buckets and you'll notice they share the same DNA. They're repetitive. They're rules-based. And they involve almost no ownership of the outcome.
If your job can be fully described in a checklist, AI is already learning it.
That's the line I keep coming back to. Not because it's clever, but because it's a useful diagnostic. Open a doc right now and try to write down exactly what you do in a checklist. If you can finish that checklist, your role is exposed. If you can't, because too much of what you do is judgment, tradeoff, persuasion, and decision-making, you're probably going to be fine. Better than fine, actually.
What AI isn't taking
I don't want this to read like a eulogy, because it isn't one. The designers I see thriving right now have a few traits in common, and none of them are about being faster in Figma.
Product designers who own outcomes.
People who define problems instead of just executing screens. People who can hold business goals, technical constraints, and user needs in their head at the same time and make a defensible call about which one wins this week. AI cannot justify that tradeoff. It can simulate the artifact, but it cannot defend the decision in a room full of people with conflicting priorities.
Designers with taste and judgment.
Taste isn't data. Judgment isn't promptable. When a founder pays a senior designer, they aren't paying for files, they're paying for confidence that this is the right move. That confidence is built on years of pattern-matching that doesn't compress into a training set easily.
Systems thinkers.
Product strategy. UX architecture. Platform-level thinking. AI is a phenomenal support for this kind of work, but it doesn't replace the person doing it. If anything, AI makes systems thinkers more valuable, because their output gets multiplied.
Communicators and leaders.
AI doesn't run meetings. It doesn't align a skeptical engineering lead with a hesitant CEO. It doesn't translate a vague founder vision into a clear brief the team can execute against. The further you move into ambiguity, the more your job looks safe.
What I'd do right now
If I were sitting in one of the at-risk roles today, here's the move.
Stop positioning yourself as fast. Speed is the worst possible thing to compete on right now, because you are guaranteed to lose. Start positioning yourself around decisions instead of deliverables. Talk about the calls you made, not the screens you produced.
Learn to direct AI instead of competing with it. The designers winning this year aren't the ones avoiding these tools, and they aren't the ones being replaced by them. They're the ones using AI to compress execution time so they can spend more of their hours on strategy, systems, and stakeholder work. That's the whole game.
Rewrite your case studies around outcomes. Not "I redesigned the dashboard," but "we shipped a redesign that moved activation by 14%, and here's why I made the calls I made." Outcomes are the language of survival from here on out.
And move upstream. Strategy beats systems. Systems beat execution. If you've been parked in execution, every hour you invest in moving up that ladder compounds.
The thing nobody wants to say out loud
The safest place in design over the next twelve months isn't being the best at making things. It's being the one who decides what gets made and why.
That's it. That's the whole post.
The designers who internalize that and adjust their work, their portfolios, and their positioning around it are going to come out of this stronger than they went in. The ones who keep insisting their value is in their Figma speed are going to have a very rough year.
I'd rather be honest with you about it now than nod along while you optimize for the wrong thing.
