The Job Description Is Wrong
Most design job descriptions being written right now are optimized for a role that is quietly disappearing. Not the designer. The specific version of the designer whose primary value was making things at a high level of craft. That version had a good run. Its successor looks quite different, and most hiring processes haven’t caught up.
I’ve been hiring designers for a while. The recent rounds have been instructive in ways I didn’t expect.
For context, I’ve written about where the role is heading. Designing Above the Interface on why execution is no longer the scarce thing, and When the Interface Assembles Itself on what the new work actually looks like in an agentic era. This article is the practical consequence of those two. You’ve redefined the role. Now you have to hire for it.
The mismatch
Candidates arrive prepared. Portfolios are polished. Case studies walk through research, ideation, iteration, delivery. The process is documented. The screens look good. And somewhere in the interview, a quiet mismatch surfaces.
The work is strong. But the underlying disposition, toward systems, toward long horizons, toward reasoning across functions, isn’t quite there. The person optimized for the job description that existed when they were last looking. That job description was written for a different era.
This isn’t a criticism of candidates. It’s a criticism of the signal we’ve been sending. We asked for Figma proficiency and pixel craft for twenty years. We got very good at producing people who have exactly that. Now the job is changing faster than the hiring process, and the gap is real.
The profile that’s losing leverage
Execution fluency is declining in value. Not to zero, craft still matters at the component level, but as a primary differentiator it is being compressed quickly. An AI can generate ten reasonable interface variations before the first standup. The person whose main contribution was producing those variations is in a structurally difficult position.
Screen-level decisions are increasingly automated. The question of which pixel goes where is less interesting than the question of what the screen is trying to do and whether that’s the right thing to be doing at all.
The designer who built a career on delivery speed, on being the person who could take a brief and produce a polished output faster than anyone else, will need to find a new edge. That edge exists. But it’s not where it used to be.
The five traits that matter now
These aren’t design skills in the traditional sense. They’re dispositions. They can come from anywhere, a design background, a product background, an engineering background. What matters is that they’re present. And the first one matters more than all the others combined.
Judgment. This is the one that doesn’t show up in any job description because it’s the hardest to screen for and the easiest to assume someone has. When an agent produces ten compositions in thirty seconds, the bottleneck immediately becomes the person deciding which one is right, why, and what to do differently with the next ten. That decision happens fast, often without complete information, and with real consequences if it’s made poorly at scale. The designer who needs three rounds of feedback and a usability test to reach a position is not built for this loop. There’s a second dimension worth naming too. Judgment about what not to build. What not to encode. Which principle is too vague to be useful and will produce incoherent outputs at scale. Knowing what to leave out requires a disciplined confidence that takes years to develop and is almost impossible to teach quickly. Everything else on this list depends on it.
Connections. Not networking. Not broad surface familiarity with adjacent functions. Genuine depth in at least one domain outside design. A designer who actually understands unit economics thinks differently about a feature decision than one who has merely attended enough finance meetings to nod convincingly. One who has shipped code, even badly, has a different relationship with feasibility. The connective tissue isn’t knowing people across the org. It’s being fluent enough in one adjacent discipline that your design thinking changes because of it. Broad literacy across five functions is comfortable and mostly decorative. One real area of depth is where the leverage is.
Conviction. When execution was the deliverable, collaboration had natural resolution points. You showed the screen. Everyone reacted. The artifact closed the loop. Now the deliverable is direction, and direction is much harder to close. There’s no moment where you show the principles document and everyone applauds. What the work requires is the capacity to shape consensus, not just participate in it. To hold a position in a room full of people who disagree, without a finished screen to hide behind. To debate without shutting down a conversation, and to know the difference between updating your view based on new information and abandoning it because someone pushed back harder. Collaboration is a given. Conviction is the rarer thing.
Long-term mindset. Roadmaps were built around what a human team could ship in a quarter. That constraint shaped how everyone thought. Remove it and most people don’t suddenly become long-term thinkers. They feel uncertain and wait for a new constraint to organize around. The ones who are energized by a longer horizon, who can reason about where a product should be in three years without a sprint to anchor to, become disproportionately valuable. But there’s a harder version of this trait worth naming. The work increasingly requires comfort with invisible authorship. Designing a system whose outputs you will never directly see, for users you will never observe, in contexts that will keep shifting. People who need to see their decisions reflected in a final artifact will struggle here. The long-term mindset isn’t just thinking further ahead. It’s being productive inside uncertainty, and being fine with never quite seeing where everything landed.
Expression and storytelling. The capacity to make an idea legible to whoever needs to act on it, in whatever form that requires. Written principles for an agent. A verbal argument in a product review. A short document that gets a skeptical engineer to understand why a constraint matters. A framing that makes a business decision feel like a design decision and vice versa. The common thread is translation. Taking something that exists clearly in your own head and making it land with someone who doesn’t share your context, your vocabulary, or your level of investment in the outcome. In the agentic era this becomes structural rather than supplementary. The design system for an AI isn’t a Figma file. It’s a set of documents. Every principle, every behavioral rule, every constraint is written down and has to be precise enough to produce consistent outputs across thousands of compositions. A designer who can only express ideas visually cannot author that system. Expression isn’t a finishing skill. It’s load-bearing.
Anyone can have these
The framing I keep encountering, that designers are well-positioned for this shift because they sit in the middle between users, business, and technology, is appealing but not quite right.
Proximity to the center is not the same as having the disposition. A designer who spent ten years optimizing screens is not automatically a systems thinker because their job description touched user research and engineering handoff. And a PM with genuine aesthetic judgment, strong cross-functional fluency, and a ten-year perspective on a market is a better hire for this work than a designer who can’t articulate a principle beyond “keep it simple.”
Things are rarely linear. The person who emerges as the author of a product’s agentic experience might come from design, from product, from engineering, occasionally from somewhere more unexpected. What they’ll have in common isn’t a title. It’s the five traits above, and the willingness to adapt to a role that is being redefined faster than any org chart can track.
The urgency is real regardless of where you’re starting from. The window to build these dispositions, to make them visible in your work and your positioning, is open. It won’t stay that way indefinitely.
If you’re hiring
Stop leading with tool requirements. Figma proficiency tells you nothing about whether someone can evaluate ten AI-generated compositions and know which one is wrong before they can explain why. It tells you they used Figma.
Test for judgment directly. Give candidates a set of outputs, design decisions, principles, compositions, and ask them to evaluate. Not explain their process. Evaluate. The ones who reach a clear position quickly, with reasoning that holds up, are showing you the most important trait on this list. The ones who hedge everything and ask for more context are showing you something too.
Look for genuine depth in one area outside design, not broad familiarity with everything. The candidate who spent two years in growth before moving into design, and who can trace the business implication of a UI decision back to an acquisition metric, is more interesting than the one who attended a lot of cross-functional meetings. Ask them to prove the depth, not just claim it.
Ask about a decision that took longer than a sprint to validate. Long-term thinkers have answers to this question that short-term ones don’t. The short-term ones will tell you about a good A/B result. The long-term ones will tell you about a direction they believed in before the data agreed, and what it cost them to hold it.
Ask them to write something in the interview. Not a case study. A principle. One sentence that could govern a design decision across a product they’ve never worked on. Expression under constraint tells you more than a polished portfolio.
And be honest about what the job actually is. If the role requires someone who can write guidelines for an agent to follow, say so. The right candidate exists. They may not be applying yet because the posting doesn’t look like it’s for them.
If you’re applying
The portfolio problem is real. Most portfolios are retrospective documents showing what you made. In the role that’s emerging, what you made is the least interesting part. What you decided, what you argued for, what you refused to let go of and why, that’s the work worth showing.
Lead with judgment, not output. A case study that shows a polished screen at the end but no evidence of the reasoning behind it is a liability in this climate. A case study that shows a messy, contested direction that you shaped and defended, even if the final artifact is unremarkable, is more honest about the skills that are becoming valuable. If you can show a moment where you evaluated quickly and were right, even better.
Build visible depth in one thing outside design. Not a bullet point on a resume claiming cross-functional experience. Actual evidence that your thinking changed because of something you understood outside the design brief. Write about it. The connections trait is hard to fake in conversation and almost impossible to hide when it’s genuine.
Practice expressing ideas in writing. Not long-form essays necessarily. Precise, short, arguable statements. Principles. Positions. The kind of sentence that could govern a decision you won’t be in the room for. If you can’t write a principle that someone else could follow without asking you a clarifying question, that’s the thing to work on.
And think about what you want to be working on in three years. Not in a career goals interview answer way. Genuinely. Because the work is opening up in ways that reward people who have a point of view about where things are going, and who are comfortable operating without a clear destination in sight. The ones still waiting for a job description that fits exactly are going to be waiting a while.
The job description is wrong. The good news is that the actual job is more interesting than the description ever was.