Figma convened more than 1,300 senior design, product, and engineering leaders at its Config Leadership Collective this June for two days of talks, panels, and peer discussions on design in the age of AI. Executives from Airbnb, OpenAI, Sephora, Atlassian, Cisco, and other companies took part. What follows are the questions the company carried into the event and the answers that landed hardest.

How do you lead a transition whose destination nobody has mapped? Ian Silber, Head of Product Design at OpenAI, frames the answer as adaptability rather than doctrine: "In just a few months our entire industry has shifted. It's hard to cling to a single process. It's more about adaptability."

Leaders are also getting their hands dirty, prototyping rough ideas and sharing attempts that fail. Airbnb Vice President of Design Teo Connor describes adopting "a beginner's mindset, where you're playing and experimenting and creating. Everyone's a beginner, and everyone should be experimenting. I've been in it with my team, prototyping myself." Room to fail matters too — Expedia Vice President of Design Rachel Been argues teams need "permission to experiment, and to be joyful about what they're creating, even when it doesn't work."

What carries over

Some practices retain their value regardless of tooling: proximity to users and commitment to craft. "A lot has changed, but the fundamentals haven't," says Sheta Chatterjee, Head of UX for Cloud AI at Google. "We're still making sure we're serving our users, that we understand their workflows. We still care about craft." Jen Dunham, former Vice President of Design at Patreon, keeps a human frame: "I show up to my team as a designer and human first. There's a high risk of chasing the tool, but fundamentally we're designing for humans, and that's not changing."

Expertise moves up the stack

Where deep skill once meant fluent execution of specific tasks, AI now absorbs a share of that work. The value shifts toward judgment. "We spent our careers taking an input, executing a task, and returning the output," says Hector Ouilhet, AWS Vice President of Design for Applied AI Solutions. "But the thing that got us into this craft isn't that. It's something higher-order: taste, discernment, judgment. That's the work only we can do." Sephora Senior Vice President of Technology Sneha Narahalli adds: "Expertise is no longer having the answer, because there will always be more questions. It's knowing which answer works best in the context you're in."

Team shape and tool adoption

Blurred roles have leaders rethinking how teams are built and who they hire. At Airbnb, Connor runs small self-contained pods that each operate like a startup, staffed with the usual product, design, and engineering mix plus perhaps a data scientist or business-side colleague — people strong in their own disciplines, bringing varied perspectives, and able to review AI output "with a strong editing eye." Dunham hires for temperament: "people who are scrappy and hungry, with wild ideas, a vocal team comfortable enough to fiercely debate one another. When everyone's nodding along with no dissent, that's a sign it's not working."

Adoption, meanwhile, needs scaffolding. Been hired a dedicated person to build the infrastructure and is standing up an education program: "In order to ask for AI usage across workstreams, we have to provide the instruction. Once you give people that security in learning, it lets them accelerate." Alex Embiricos, Head of Enterprise at OpenAI, recommends starting small rather than mapping large workflows for automation — ask an agent to summarize one unwanted Slack thread, and familiarity spreads from there.

Earning trust in agents

Agents behave inconsistently, which complicates both building and trusting them. Two recurring principles: introduce autonomy in stages, and keep interfaces familiar. Chatterjee builds checkpoints in by design — "We don't hand agents full autonomy right away. We start them on small tasks with approval steps built in" — and removes them as confidence grows. Atlassian Chief Design Officer Charlie Sutton warns against overreaching: "A lot of good design is about balancing the familiar and the novel." Customers, he says, "often need those handholds and that familiarity."

Holding the quality bar

When shipping is cheap, deciding what deserves to ship becomes the differentiator. Silber places responsibility with leadership: "As leaders, you have to partner with other leaders to hold the bar from the top. When it looks like leadership just cares about moving fast, that's a bad signal." Connor's team uses an old-school method, walking one another through decisions to build "a collective understanding of what good looks like." And Cisco President and Chief Product Officer Jeetu Patel returns to human judgment: "You can't outsource taste. As you start building code automatically, what matters most becomes taste and good judgment. That's not something you can hand to an agent."

The throughline: leaders are working these questions hands-on — building, experimenting, comparing notes, and admitting what remains unsettled — while treating taste, judgment, curiosity, and care as no less essential than before.