For the better part of a decade, the T-shaped designer was the hiring ideal.
The model was elegant and intuitive: deep expertise in one discipline — visual design, interaction design, research — combined with enough breadth across adjacent areas to collaborate fluidly with engineers, product managers, and business stakeholders. The T captured something true about what made designers effective in cross-functional teams. It became the template for job descriptions, career ladders, and interview processes across the industry.
It was a good model. And it was built for a world that is ending.
The T-shape assumes there is a meaningful execution layer for the human to own. You go deep to do the craft work — the designing, the building, the documenting, the delivering. You go broad to collaborate around that craft work. The shape points, in both directions, toward making.
AI is hollowing out the making layer. Not eventually. Now. And the implications for what we should be hiring for are significant, immediate, and almost entirely unaddressed in how most design organizations are actually recruiting.
To understand why the T-shape is breaking, it helps to be precise about what it was actually optimizing for.
The depth of the T was about craft mastery — the ability to produce excellent work in a specific domain. Visual hierarchy, interaction patterns, research methodology, design systems architecture. These are skills developed through deliberate practice over years. The depth represented a real and hard-won capability.
The breadth of the T was about collaboration surface — the ability to communicate and coordinate across disciplinary boundaries without losing the thread of your own expertise. A visual designer who understood enough about information architecture to have a productive conversation with an IA. A researcher who understood enough about visual design to give feedback that was actually useful.
Both of these things still matter. But the ratio between them is shifting. And more importantly, the nature of what depth and breadth mean is changing fundamentally.
When AI can produce competent first-pass work across most of the execution layer — when it can generate visual directions, draft interaction patterns, synthesize research themes, and produce component variations at a pace and volume no individual human can match — the value of human depth in any single execution domain decreases. Not to zero. But it decreases.
What increases in value is something the T-shape doesn't have a clean axis for: the ability to work across the AI's outputs with speed, judgment, and intent. To be the directing intelligence rather than the executing one.
In an AI-augmented design environment, the work moves through four stages. Each stage requires a distinct capability. Together they describe a different shape of designer than the T captured.
Stage 1: Curation — before the machine runs
The first stage is about intent and constraint. Before any AI tool generates anything, someone has to decide what it should be trying to produce. What problem is being solved. What constraints apply. What success looks like. What aesthetic and functional parameters should shape the output. What shouldn't be attempted.
This is curatorial work, and it requires a specific kind of judgment: the ability to translate a vague brief or a felt sense of what's needed into precise enough direction that a system can act on it usefully. Vague intent produces vague output. The designer who can front-load their taste — who can encode their judgment into the inputs before the machine runs — produces dramatically better results than the one who generates first and edits later.
This stage also requires constraint intelligence. Knowing what not to ask for is as important as knowing what to ask for. The ability to define the edges of a problem sharply enough that the system operates within them is a real skill. It is not a skill that design education has historically prioritized.
Stage 2: Orchestration — managing what the machine does
The second stage is about coordination across tools, agents, and workflows. As AI capabilities become more agentic — able to execute multi-step tasks, operate across platforms, and produce outputs that feed into other processes — the designer's role increasingly involves directing and managing these workflows rather than executing the individual steps within them.
This is orchestration. And it is genuinely new territory for most designers.
Orchestration requires systems thinking at a level of abstraction that goes beyond designing a system to designing the process that produces the system. It requires enough technical literacy to understand what the tools can and cannot do, where the seams are, and where human judgment needs to be inserted into the workflow rather than assumed at the end of it. It requires comfort with ambiguity at the process level — the ability to manage a workflow that is itself evolving — rather than just at the design problem level.
The designers who are good at orchestration tend to be the ones who were always slightly more interested in how things work than in how things look. They are not necessarily the ones with the most impressive portfolios.
Stage 3: Curation — after the machine returns
The third stage brings curation back, but in a different mode. Where the first curation was about shaping inputs, this one is about evaluating outputs. The machine has run. It has returned a volume of options — directions, iterations, variations, drafts. Now someone has to sort them.
This is where earned taste does its most critical work in the new model. The ability to move quickly through a large set of options, identify what is genuinely promising, recognize what is superficially attractive but structurally flawed, and make confident decisions about what to develop further and what to discard — this is a skill that requires both speed and depth of judgment.
It also requires a specific kind of psychological resilience. The volume AI produces can be overwhelming. The options that look good on first pass are not always the ones that hold up under scrutiny. The ability to maintain a clear sense of what you are looking for — and to not be seduced by novelty or surface quality — is harder than it sounds when you are working through hundreds of generated directions under time pressure.
Stage 4: Presentation — making the output land
The fourth stage is the one that remains most irreducibly human for now. Not because AI cannot produce a presentation, but because the room still wants to hear it from a person. The trust transfer, the reading of the audience, the ability to defend a decision under pressure and adapt in real time to the questions being asked — this is still human territory.
But presentation in this model is doing different work than it used to. The designer is no longer presenting their own creative work. They are presenting curated, orchestrated, AI-augmented output — and the challenge is to own it with the same authority and conviction as if they had made every pixel themselves. The ability to stand behind work that was produced through a fundamentally different process, and to articulate the judgment and intent that shaped it, is a new communication skill that the field hasn't fully reckoned with.
The gap between the new shape of valuable design work and the way most organizations are hiring for it is significant. And it is largely structural.
Job descriptions are written by humans who learned design in the old model, approved by HR systems that haven't been updated, and evaluated against interview frameworks that were designed to assess execution-layer skills. They ask for Figma proficiency, portfolio depth, process documentation, and craft capability. These are not irrelevant. But they are describing the old job.
The new job requires something different and something harder to evaluate in a standard interview process: the judgment to constrain a system before it generates, the intelligence to orchestrate across tools and workflows, the discernment to evaluate volume at speed, and the communication ability to make AI-augmented output land with people who don't care how it was made.
None of these things show up in a portfolio. Most of them don't show up in a traditional interview. They are almost entirely invisible to the standard hiring process — which means organizations that haven't updated their evaluation frameworks are not finding these people, even when they are in the pipeline.
There is also a cultural gap. The designers who are best positioned for the new model are not always the ones who present as the most conventionally excellent designers. They may not have the most polished case studies. They may be harder to evaluate quickly. They may describe their work in ways that sound less like the design culture most hiring managers were trained in. Recognizing them requires a different lens — and developing that lens requires acknowledging that the old lens is increasingly optimized for the wrong thing.
This is not a call to stop hiring designers with strong craft skills. Craft still matters. Earned taste still matters. The argument is not that execution capability is irrelevant — it's that execution capability alone is no longer sufficient, and that the additional capabilities that now matter most are not what most hiring processes are designed to surface.
The designers worth finding right now are the ones who combine experiential judgment with adaptability — who have built real taste through real work, and who have demonstrated the intellectual flexibility to operate in new modalities rather than waiting for the modalities to stabilize before engaging with them.
In practical terms, that means looking for:
Evidence of curatorial intelligence. Can they articulate constraints before they start working? Can they describe a problem precisely enough to direct a collaborator — human or machine — without over-specifying? Do they ask the right questions before diving in?
Comfort with orchestration. Are they curious about how tools and systems work, not just what they produce? Have they developed their own workflows, their own ways of combining capabilities to get to better outcomes faster? Are they building with AI or just using it?
Speed and confidence in evaluation. When presented with a volume of options, can they move through them quickly and make defensible decisions? Can they articulate not just what they chose but why — and why the alternatives didn't make the cut?
Communication that owns the work. Can they present AI-augmented output with the same authority and conviction as work they made entirely themselves? Can they defend curatorial decisions under pressure in a room that doesn't understand or care about the process?
These are not exotic requirements. They are the natural evolution of what great design judgment has always required — applied to a new context. The designers who have them exist. Finding them requires updating how you look.
There is a certain irony in design organizations — functions that exist to improve the experience of using things — running hiring processes that are poorly designed for what they are actually trying to accomplish.
The experience of interviewing for a senior design role at most organizations is optimized for the interviewer's convenience, not for surfacing the signals that actually predict performance. Long take-home exercises that test endurance more than judgment. Portfolio reviews that evaluate presentation skill more than design thinking. Interview panels that ask the same questions across five rounds and mistake consistency for rigor.
If the shape of the valuable designer is changing, the shape of the hiring process needs to change with it. That means replacing long exercises with live challenges that test judgment under pressure. It means adding evaluation moments that are specifically designed to surface orchestration capability and curatorial speed. It means including non-designers in the process to see how candidates communicate across disciplinary boundaries. It means updating job descriptions to describe the actual job, not the job from three years ago.
None of this is difficult in principle. It requires acknowledging that the old process is producing increasingly unreliable results — and being willing to build something better.
If you lead a design organization, the question this piece is asking you to sit with is simple: does your hiring process find the designers the next 24 months will require?
Not the designers who are excellent at the job as it existed. The designers who are positioned for the job as it is becoming.
The answer, for most organizations, is no. Not because the people aren't there, but because the process isn't looking for them in the right way.
That gap is closable. But closing it requires moving before it's obvious — which is exactly the kind of decision that distinguishes leaders who shape what's coming from leaders who react to it after the fact.