How to build a design team in the AI era.

The design team I would build today looks almost nothing like the one I would have built in 2020. Same people, in many cases. Same talent. But the shape is different, and the shape is the whole game now.

For most of my 15 years leading and doing design, the smart way to build a team was a production line. You hired specialists. A researcher who only did research. A UX designer who handed off to a visual designer who handed off to a prototyper who handed off to an engineer. Each person was deep in one craft, and the work moved down the line. It felt efficient. It looked like a real org chart. For a long time it was the right answer.

It is the wrong answer now. AI made the making fast. And when making gets fast, the handoffs become the slowest thing in the building.

The handoff is the bottleneck now

Think about where time actually went on the old production line. A little bit of it went to the actual work. Most of it went to the gaps between people. Writing the research brief. Waiting for the readout. Translating the readout into a spec. Waiting for the design. Reviewing the design. Sending notes back. Waiting again. The work itself was never the slow part. The waiting was.

Every handoff is a tax. Context gets dropped at the seam. Intent gets watered down with each pass. The researcher knows something the designer never quite hears. The designer makes a choice the engineer never quite understands. Three smart people, and the idea loses a little of itself at every border crossing.

AI did not shrink the work. It shrank everything around the work. A first draft of a flow takes minutes. A working prototype takes an afternoon, not a sprint. Twenty copy options, ten layout directions, a styled front-end you can actually click. All cheap now. So the math flips. When making is cheap and handoffs are expensive, the team that wins is the team with the fewest handoffs.

That single idea reorganizes everything. You stop optimizing for deep specialists who pass work along. You start optimizing for whole humans who carry an idea from question to shipped.

Hire for range, not just depth

Here is the bet I make now. One designer who can run their own research, build their own prototype, and ship their own front-end beats three specialists passing work between them. Not because the one person is better at each craft than each specialist. They usually are not. They win because nothing falls in the gaps, because there are no gaps.

This is the biggest shift in how I hire. I used to look for the deepest point on someone’s spike. The best pure visual designer. The sharpest researcher. Now I look for range. I want the person who gets curious past the edge of their lane. The designer who taught themselves to run a usability test because they were tired of waiting for one. The one who learned enough front-end to ship the thing instead of describing it. The one who writes the copy instead of leaving a gray box that says text here.

Range does not mean shallow. That is the mistake people make when they hear this. I am not asking for someone who is mediocre at six things. I am asking for someone with real depth in one or two crafts and genuine working fluency in the rest. Deep enough to have taste. Broad enough to never get stuck waiting. AI fills in the floor under the skills they have not mastered, so a designer with strong judgment and decent range now operates like a whole team used to.

I wrote a longer piece on the fundamentals of team building that still hold, the parts AI did not change, in my notes on how to build a design team. The org-chart logic is what shifted. The human logic underneath it did not. You still hire for curiosity, ownership, and judgment. You just weight range higher than you used to, and you stop rewarding people for guarding a narrow lane.

If you want the specifics of how I screen for this, I put it in what I look for when I hire. The short version is that I care more about how someone got unstuck than about the polish of their best screen. Range shows up in the stories about the messy middle.

Make AI the default, not the exception

Most teams get this part backwards. They treat AI as a special occasion. There is a tool somewhere, a few people use it, and everyone agrees it is interesting. It sits to the side of the real work, like a toy on a shelf. Nothing changes, because the path people walk every day did not change.

Behavior follows the path of least resistance. If using AI is a detour, people will not take it. They will take it once, feel clever, and quietly go back to the way they have always worked. The only way it sticks is if you build it into the path people already walk. It has to be the easy way, the default way, the thing that takes less effort than not using it.

That means the AI lives inside the actual workflow. It is in the brief template. It is in the prototype starter. It is in the review step. The new hire does not get a class on prompting. They watch how the work moves and they copy it, and the AI is already woven into how the work moves. You make the good behavior the lazy behavior, and then it happens on its own.

In a multi-billion-dollar design org I led, this is exactly how the real wins showed up. We did not hand people a tool and hope. We rebuilt the path. I built an agent that took one painful internal workflow from around 320 hours down to about one hour. The number is not the point. The point is what it freed up. Nobody on that team was nostalgic for the 320 hours. They got to spend that time on judgment instead of grind.

I laid out the mechanics of this in the AI-first process this runs on, because the structure matters more than the specific tools. Tools change every few months. The principle, that AI belongs inside the default path and not beside it, does not.

When making is cheap, taste is the job

Here is the part nobody wants to hear. When anyone can make almost anything in minutes, the making stops being valuable. The judgment about what is worth making becomes the whole job. Taste is no longer the soft skill. It is the hard skill, the scarce one, the thing you cannot fake or generate.

AI will happily make you a thousand options. It has no opinion about which one is right. It does not know your customer is a tired parent at 9pm. It does not know that the third option is technically prettier but emotionally cold. It does not feel the difference between fine and right. A human does. That feel, built over years of shipping real things to real people, is the part that does not get cheaper. If anything it gets more valuable, because there is so much more output to judge.

So the leader’s job changes shape. I used to be one of the best makers in the room, and a lot of my value was in the artifact I could produce. That is not where my value is now. Now I defend the bar. I am the person who looks at forty good-enough options and says no to thirty-nine of them. I protect the standard when the easy thing is to ship the first thing that clears the wall.

Defending taste is harder than making, and it is lonelier. When making was slow, the bar defended itself, because nobody had time to produce junk. Now junk is free and abundant and arrives faster than you can review it. The pressure to just ship it is constant. A leader who does not hold the line will watch their product fill up with things that are fine and nothing that is great. Fine is the enemy now. Fine is what the machine makes by default.

What this looks like at Story Genie

I live inside this every day at Story Genie, the company I founded. Agents write and illustrate personalized hardcover kids’ books. A child becomes the hero of their own story, and the whole thing comes together in about 60 seconds for a preview, with a real human in the loop supervising before anything gets printed and shipped.

On the surface that sounds like a story about automation replacing people. It is the opposite. The agents made the making nearly free. A custom illustrated book used to take an artist days. Now the draft exists in a minute. So what is left for the humans? Everything that matters. Is this story actually good. Does this illustration feel warm or does it feel like a machine guessed. Would a parent be proud to put this on the shelf. We let people see the full book before they pay a cent, because we are confident in the judgment behind it, not just the speed in front of it.

The team that builds this cannot be a production line of specialists waiting on each other. It has to be a handful of people with range and taste, with the AI woven into the default path, all defending a bar that the machine will never defend on its own. That is the whole thesis of this article, running in a real company, shipping real books to real families.

Build the team for the speed you actually have

The hard truth is that most design orgs are still shaped for a world that ended. They are still hiring deep specialists, still running work down a handoff line, still treating AI as a side project, still rewarding the best maker instead of the best judge. They are building a 2020 team in a year that moves at a different speed, and they wonder why they feel slow.

You do not fix that with a pep talk about innovation. You fix it with structure. Hire for range. Kill the handoffs. Put the AI on the default path. Promote the people with taste and let them defend the bar without apology. The culture you want follows from the way the work actually moves, not from the poster on the wall.

Start with the team you have, not the one you wish you had

Most advice about building a design team in the AI era assumes you get to start fresh. You almost never do. You inherit an org chart full of specialists who were great hires for the world you hired them into. The illustrator who only illustrates. The researcher who only researches. The motion designer who lives in one tool. They are not the problem. They are talented people sitting inside a shape that no longer fits. Your job is to reshape the work without making good people feel like they are being thrown away.

The first thing I tell every specialist is the truth. Their craft is not being deleted. It is being unbundled. The deep skill they built is still real and still rare. What is changing is that the deep skill alone no longer fills a whole role, because the slow parts around it just got cheap. A great illustrator used to spend most of the week making the thing. Now the making is fast, and the rare part is knowing what is worth making and why. That is still them. It is just the better half of them.

Some people hear that and lean in. They are the ones who were always a little bored by the repetitive part of their craft anyway. They want to touch more of the problem. Give them range early and they sprint. Others hear the same words and feel the floor move, because their identity is wrapped tight around being the best at one narrow thing. Both reactions are normal. Watch for them, name them out loud, and do not punish the second group for being honest about fear. Fear is not resistance. Fear is someone who cares about being good telling you they are not sure they can be good at the new thing yet.

Real resistance is different, and you will know it when you see it. It sounds like proving the new tools wrong instead of learning them. It sounds like protecting a handoff that only exists to protect a job. A few people will dig in no matter how much room you give them, and at some point you have to be honest that the role they want no longer exists here. But that is a small group, and you earn the right to that conversation only after you have genuinely tried to bring them along. Most people do not resist. They wait to see if you mean it.

So mean it. Retrain toward judgment, not toward more tools. Tools are the easy part and they change every quarter. The thing to grow is taste, the read on what is good, the instinct for what the work is actually for. Pair your deep specialist with a broad problem and let them feel how their eye carries across it. Have them direct the AI instead of competing with it, critique ten cheap options instead of laboring over one. The skill they trusted is exactly the skill that makes them a great director. They just never had enough volume to use it that way before.

Lead all of this as growth, because it is. This is not a demotion of their craft. It is a promotion past the grind that used to bury it. Frame it that way in every one on one and mean every word. The illustrator becomes an art director with a faster brush. The researcher becomes the person who decides what is worth building, not just what users said. When you tell a specialist that the rarest, most human part of their work just became the whole job, most of them light up. They were always more than the task. AI just gave you the excuse to finally use all of them.

When making is free, the only thing left to lead is judgment. So build the team around the judges.

None of this is theory for me. It is how I have built and run design at companies like Kinjo, Supply Drop, The Mysterious Package Company, Surf Studios, and Enverus, and how I run Story Genie right now. The shape changed. The mission did not. If you are trying to reshape your own team for this, I am happy to help you build it.

You can reach me through the page where I offer to help you build it, and we can figure out what the right shape looks like for your team.