Design Teams in the Agentic Era

My thoughts on what comes next

The build side of product development just got 10x faster. The thinking side didn’t.

That’s it. That’s the whole story. Someone on your team is running seven agents simultaneously, shipping at a pace that makes your upstream process look like it’s standing still. And yet we still need to figure out what we’re solving, align with product and engineering, wrestle ambiguity into something buildable. That work didn’t change. The math around it did.

All this means the designer’s job changed. Not disappeared — changed. And I think it changed for the better. Here’s why.

---

The New Job

I lead a large design organization at Cisco — Security, AI, and Platform design. I’ve been watching this shift play out across my teams in real time. And what I keep seeing is that the design teams thriving right now share three priorities. None of them are optional.

1. Set direction. Your most valuable output is no longer a Figma file. It’s clarity about where you’re going and why. When agents can generate ten options in an hour, the person who can look at all ten and say “none of these — here’s why” becomes the most important person on the team. Direction is the multiplier. Without it, speed is just chaos.
   
   I’ve been playing with something I’m calling the Agentic Pod — a multi-agent pipeline where AI agents play the roles of PM, researcher, designer, and engineer, each producing real artifacts that build on the last. The thing that surprised me most is how badly it fails without clear direction at the start. Garbage in, garbage out — at 10x speed. The human who frames the problem well is the difference between a useful pipeline and an expensive random walk.
   

2. Own the system. Design owns Magnetic which is Cisco’s design system for admin experiences. And I’m now convinced that design systems are about to become the single most important investment a design org can make.
   
   Last week, one of my design directors pointed Claude Code at Magnetic and asked it to build a security detection prototype. Real components, real navigation, theme switching, working admin panels — running in ten minutes. Then he connected it to our research repository and it built 44 detection detail panels, every design decision tracing back to something a real customer said. That happened because the AI had access to our design system. Every component it pulled was ours. Every pattern matched our standards. The design system was the design review.
   
   But here’s the bigger point: the design system needs to evolve. It’s not enough to give agents access to your component library. The system needs to include the skills we want any coding agent to pick up — our design principles, our philosophy, a codification of our design strategy so that the agent starts making decisions that align with what’s important to us as a design org. For many organizations, they’ll scramble to even articulate what these skills should codify. But the exercise itself is incredibly valuable. It forces you to make explicit what was always implicit about your design culture.
   
   Your design system is your leverage. It’s how your taste scales.
   
   The teams that invest here will see their design decisions show up in every agent-generated output, automatically. The teams that don’t will spend all their time cleaning up messes that a good system would have prevented.
   

3. Stay close to customers. This is the one AI can’t touch, and it’s the one I feel most strongly about.
   
   AI can summarize a call transcript. It can pull themes from a hundred interviews. But it cannot build the kind of understanding you get from sitting across from someone while they use your product. Watching where they hesitate. Where they light up. Where they give up. That’s not data — that’s intuition built through proximity. And it’s irreplaceable.
   
   The designers and researchers who thrive in this world will spend more time with users, not less. Because when the build is fast, knowing what to build is the only competitive advantage left.

---

A Word About Process

I’m not going to tell you the design process is dead. That framing is hyperbolic, and it misses the point.

What I will say is this: the design process was always a scaffold. The double diamond, the sprints, the handoff rituals — they were useful structures for managing the fact that building things was slow and expensive. They sequenced the work because the work needed to be sequenced when each phase took weeks.

That constraint is loosening. When someone can prototype in minutes, the idea that design always happens upstream of engineering stops making physical sense. The phases blur. Design becomes something that happens throughout, not before.

Here’s the uncomfortable part: this requires us to operate at a velocity that will feel wrong. Good design requires thought, strategy, research, experimentation. We spent years shifting left in the process — getting ahead of engineering so we’d have the time and space to do these things well. AI just ate our buffer. Once again we’ll have engineering pulling work from us faster than we can think through it.

So we need to figure out how AI helps us think faster. Process faster. Experiment faster. Not just how it helps us build faster — that’s the easy part. The hard part is compressing the messy upstream work without losing what makes it valuable.

A designer who’s embedded throughout the build — shaping direction at the start, consulting during, polishing craft at the end — has more influence than one sitting upstream producing documents that get thrown over a wall. But it requires different skills. You need to get comfortable with code and the systems that manage it. Give feedback on implementations, not mockups. Work in the medium, not above it. And you need to trust that the first version won’t be perfect and that your job is to make it better, not prevent it from existing.

---

Leaders: Get Your Hands Dirty

You cannot lead a transformation you haven’t lived. And I don’t mean “watched a demo.” I don’t mean “read the blog post.” I mean: can you sit down with the tools your team uses and actually do the work?

I spent the last year building AI agents myself. Not managing people who build them — building them. I built an Agentic Pod. I built a personal AI assistant that manages my daily workflow. I connected Claude Code to our design system and our research corpus and watched what happened.

And here’s what I learned that I couldn’t have learned from a distance: the bottleneck really does shift from production to judgment. When the build is fast, the quality of your direction becomes the thing that matters most. Bad direction at AI speed is worse than bad direction at human speed, because you get more of it, faster.

The leaders who manage this shift from the conference room — treating AI as a strategy slide instead of a lived practice — will lose their teams’ trust. Your people know the difference between someone who’s used the tools and someone who’s read about them.

If you can’t use the tools your team uses, that’s your first priority. Not next quarter. Now.

---

What I Believe

I want to end with where I’ve landed. Not a framework. Not a prediction. Just the convictions I’m building on — with full acknowledgment that this space is moving so fast, some of this will be stale by the time you read it. That’s the deal now.

Design teams aren’t shrinking. They’re stretching — into systems, into code, into direction-setting that was always their real superpower but often got buried under production work. The freed capacity goes to the stuff that actually matters: customers, decisions, quality.

But let’s be honest about something: there are a lot of designers out there whose value lives almost entirely in production. Designers who are frustrated by ambiguity, frustrated by the lack of clear direction from product management. The ones who want to be told what to build and then build it beautifully. Those designers are the ones who will feel this transformation the most. Not because they’re bad at their jobs — but because AI just got very good at theirs.

Agents don’t replace taste. They make taste scale. A great design system plus AI agents means your best decisions propagate everywhere, automatically. That’s unprecedented leverage for design organizations.

The best design leaders right now are the ones willing to be beginners again. To put their hands on the tools. To feel the discomfort of not knowing and push through it anyway.

Direction compounds. Gatekeeping decays.

The teams that shift from controlling outputs to setting direction will have more influence than they’ve ever had. The ones that cling to the old sequence will become a bottleneck that gets routed around.

This is the best time to lead a design team in a decade. I genuinely believe that. Not because it’s easy — it’s uncomfortable and uncertain and nobody has all the answers. But because for the first time, the constraint isn’t production capacity. The constraint is vision. And vision is what design has always been about.

The question is whether you’ll lead the change or get pulled along by it.

I know which one I’m choosing.

Jason Cyr is VP and Head of Design for Cisco Security and AI, leading a team of 200+ across security, AI, and platform design. He writes about design leadership and AI agents here on Substack, LinkedIn and YouTube.