Six months ago I pulled my lead designer into a conference room and told her I needed to change her job description. She already knew. She had been waiting for me to catch up.

The conversation every design manager is having right now goes something like this. We still need people, right? And the honest answer is yes, but probably not for the reasons you think.

Here is what actually changed. The old model was a funnel. Research feeds strategy, strategy feeds design, design feeds engineering. Every handoff was a translation, and every translation was a tax. You lost something each time. What the new generation of AI tools does is collapse that funnel. You can go from a prompt to a functional prototype in a single afternoon. I have seen it happen. The bottleneck is no longer production. It is judgment.

And judgment is the thing you cannot automate.

The most valuable thing a designer does right now is sit in front of a prototype the AI built and say: this is wrong. And then explain why. Not gut feeling wrong. Specifically wrong, in a way that can be acted on. That takes taste. It takes context. It takes a working model of the person who is going to use the thing. None of that is in the model.

I have been experimenting with Stitch for UI generation. My team has too. Workflows that used to take three days now take an afternoon. That is not a threat. That is a gift, if you redirect the time toward the things only humans can do. The mistake is treating it as a shortcut to shipping. It is a shortcut to learning. Faster feedback, faster death for bad ideas, more room to try things you would never have had the budget to explore before.

Which brings me to the question nobody wants to answer out loud: who presents the work when the robot made it?

My instinct was always to put the most senior person in the room. The right answer is whoever can defend the decisions, not just the output. Sometimes that is the same person. Sometimes it is a PM who actually understands what the model was optimizing for. The thing you cannot do is have someone present work they cannot explain. We started requiring a named decision owner on every AI-assisted design. Not the person who typed the prompt. The person who is accountable for what ships.

Design systems matter more now, not less. This surprised me. When AI can generate UI on demand, why bother maintaining a system? Because the system is no longer just a component library. It is the vocabulary that lets you evaluate AI output at speed. Without it, every generated screen is a one-off. With it, you have a standard to hold things to. Your systems designer is one of the most important people in your org right now. If you do not have one, that is the gap.

Can you collapse research and product design into a single function? I think yes. With conditions.

The reason they were separate was throughput. Researchers moved slower than designers, so the org adapted by splitting them. AI has changed the throughput equation on the design side. The distance between the two functions can close. But only if your designers can actually do research. Not recruiting and scheduling. I mean forming hypotheses, reading behavioral data, synthesizing a finding, holding it against a prototype. That is a learnable skill. Most designers just have not been given the expectation.

The new job description for a senior designer looks something like this. Own the insight. Own the artifact. Own the room when the work gets presented. Own the outcome. If they cannot do all four, they are a junior designer with a senior title.

The question of where the human is most useful has a clear answer. Not in production. Not in generation. In the three moments that actually require judgment. Before the work starts, when you decide what problem you are solving. When the output lands and someone has to say whether it is right. When the work ships and you find out if it did what you thought.

Every other step can be augmented. Those three cannot.

The org chart you had in 2022 is not the one you need now. The only real question is whether you change it on purpose or whether it changes you.

Josh

Keep Reading