When I tell people I have a master’s degree in Design Management, I usually get one of two reactions. The first is polite confusion. The second is the assumption that I studied how to manage designers.
The second guess isn’t entirely wrong. Managing creative teams is part of it. But the actual discipline is a lot wider than running projects or making sure the logo gets used correctly. So here’s the honest version. Somebody has to know what we’re making, why it matters, and how all of it stays coherent once a lot of people (and now a lot of systems) are touching it. That’s the whole discipline. Design Management is just the name it got. And it’s the kind of work that’s basically invisible when it’s going well. You notice it when it’s missing: a product that feels disjointed, a brand that contradicts itself from one channel to the next, a clever feature solving a problem nobody actually had.
For most of its history, that skill set lived off to the side somewhere between brand strategy, product design, and organizational leadership. AI is pulling it toward the center. As models, agents, and generative interfaces change how software gets built, managing design as a system stops being a nice-to-have and becomes the thing that decides whether the AI actually works for people.
This isn’t an argument for the degree, by the way. The credential gave me a name and a structure for a way of thinking, but the thinking doesn’t belong to a diploma. Plenty of people already practice it without calling it anything: product managers who keep the whole system in view, founders who can hold strategy and craft in the same thought. The point is the skill set. That part is open to anyone willing to work this way.
AI changes the structure of experience itself
Most of the current conversation about AI is about tools. Which model is best, which platform is fastest, which app can spin up copy or code or a slide deck in seconds. Fair questions. They’re also the smallest part of what’s changing.
The deeper shift is that AI is changing the structure of the experience itself. We’re moving from static interfaces to adaptive ones, from fixed flows to systems that read intent, generate options, and reshape the interface while you’re still looking at it. And once an experience is generated instead of drawn screen by screen, the central design question moves upstream. It stops being “what should this button look like” and becomes “what rules and context should guide the system that produces the button (and everything around it) for this specific person, in this moment.”
Think about a checkout page. A designed checkout page is a fixed object. You can point at it. A generated checkout experience is a set of decisions a system makes in real time, and somebody has to decide what those decisions should be. It should probably behave one way for a returning customer buying a single familiar item, and a different way for a first-time visitor comparing three options and hesitating over the shipping cost.
The value is everything around the model
This is where Design Management earns its keep. A model will happily generate a layout, write a headline, or produce working code. It’ll just as happily produce something generic, off-brand, or strategically hollow. The prompt is only the surface. The real work is everything the model draws on: the design system it builds from, the brand rules it has to respect, the criteria its output gets measured against, and the judgment call about when it should generate, when it should only assist, and when it should stay out of the way entirely.
And that work is the designer’s job now. Not a separate management layer floating above it. Once a tool lets anyone crank out more, faster, the cranking stops being your advantage. The floor rises, and what’s left is deciding what’s worth producing at all.
GenUI needs more than generation
Generative UI points toward a world where interfaces aren’t fully decided in advance. Parts of the experience assemble themselves around your goal, your context, your prompt. Which is genuinely exciting. It also introduces a whole new category of design problem.
Because when the interface itself can change, the hard problems are about what DOESN’T. An experience whose layout differs every time still has to stay usable and coherent. A person who can ask for almost anything still has to feel guided instead of lost. And underneath all of that sits the question that actually matters… when the model produces the interface, who answers for whether it’s any good?
I saw a version of this long before AI was involved. On a client pitch, the team had every right ingredient: smart strategy, strong visuals, technical depth, relevant case studies, a real business opportunity. But each piece had been built from a slightly different angle, so the whole thing pulled in several directions at once. One section sold innovation. The next sold delivery confidence. Then technical expertise, then brand experience. None of it was wrong. The audience just had to do too much work to figure out what mattered most.
It didn’t fail dramatically. It dragged. Extra meetings, repeated revisions, subjective feedback, a deck reworked until the argument finally read as one connected experience instead of a pile of good parts.
That’s what AI makes urgent. If a team can fragment a story one slide at a time, a model can do it at scale.
Design Management has an actual answer here. Quality stops living in the final screen, because there isn’t a single final screen anymore. It moves upstream, into the system that produces the screens. Which means it belongs to whoever set the standard the output has to meet. Generating the thing is the easy part. Designing the conditions where generation reliably produces something worth shipping… that’s the work.
Accessibility makes this concrete. You can audit a fixed screen once for color contrast and screen-reader support and trust that it holds (though most of the web fails even that check: 94.8% of the top million home pages have detectable WCAG errors). A generated interface doesn’t get that luxury. It has to meet the same standard on every variation it assembles, for every user, with nobody checking each one by hand. You can’t prompt that guarantee into existence after the fact. It has to be built into the system as a rule the generation is never allowed to break.
AI makes taste and judgment more valuable
There’s a real fear that AI will flatten creative work. That everything will start to look, sound, and feel the same. The fear isn’t unfounded. A model is very good at producing the average of everything that already exists, and it can hand you polished, plausible work almost instantly. That’s not a flaw somebody will patch out. It’s how the technology works. Left alone, a model drifts toward the familiar, and polished isn’t the same as meaningful.
Which is exactly where human judgment matters most. Someone still has to know what’s actually good, what’s true to the brand, and when an output is technically correct but strategically wrong. If you want something better than average out of a model, it takes a person who knows what “better” looks like and can steer toward it on purpose. Taste stops being decoration and becomes a working filter on everything the system produces.
The scarce skill is coherence
The longer I work around AI, the more obvious this gets: the opportunity isn’t in mastering any single tool. The tools keep changing (so do the models, and so do the interfaces built on them). What gets scarce is coherence. The ability to make a flood of fast, cheap, generated parts add up to something that actually serves a person.
That’s the work AI raises rather than removes. As models get more capable, the job shifts away from making every individual thing by hand and toward setting the standards and strategy that decide what gets made and whether it’s any good. That’s management in the best sense of the word. It points creativity at a purpose instead of controlling it.
The name on my degree still earns polite confusion. The work it names just became the job.