The Morning Download: Figma CEO Says Design is the Competitive Edge in Age of AI

Happening now. Global semiconductor stocks skidded after the U.S.-China summit concluded without major chip deals. No breakthrough on Nvidia sales to China put a crimp in the AI chipmaker's seven-day, 20% surge, but analysts see the dip as temporary given next Wednesday's earnings report.

Also...AI chip company Cerebras Systems saw its shares soar 68% in its first day of trading on Thursday.

Highlights from Barron’s coverage:

• Figma exceeded first quarter expectations with earnings of 10 cents a share and a 46% gain in revenue to $333.4 million. It raised full-year guidance.
• Paid customers grew 54% to 690,000, with “pro team conversion” up 150% due to AI feature adoption.
• Despite an 8% after-hours stock jump, Figma shares are down 46% this year due to AI disruption fears and new competitor tools.

And as Barron’s notes, Figma announced a partnership in February with Anthropic that lets users turn AI-generated code into designs. It also has agreements with OpenAI and Google.

I spoke to Field yesterday afternoon about the role of AI in Figma’s design platform and his approach to leading the company through the extremely volatile period since its July stock market debut, when shares soared 250%. He said the company is in a unique position to fuse AI with the unverifiable judgement calls and creativity where humans excel.

WSJ Leadership Institute: Well, you wanted to go public!  Shares are down since the IPO. Have you reached a turning point?

Field: I think it all just reinforces, along with our raise on the guidance side, that Figma is where the world’s best software teams are building the world’s best software. And the other thing is we're not in the place where, you know, we're replacing our engineers with AI agents, but it sure is the case that AI and LLMs are rapidly accelerating. The capabilities are incredible and I think that we can all project forward to a future where code is commoditized. And as that happens, design is the differentiation layer, the competitive edge.

I’ve learned to not speculate on the forces of the market. The only thing that we can do is deliver strong results and make sure that we have the right strategy for the long term, to build a great business.

I actually think that the team is now in a place where we're accelerating so rapidly that they are surpassing my expectations and coming up with things that I haven't even said we’ve got to go hit that goal. The innovation has been wild and the creativity as well.

WSJLI: How is the role of human judgement changing as AI becomes more capable?

Field: The best human progress usually [draws on] things that are not verifiable. And design is inherently not verifiable. And so as people are really trying to push the edges and the boundaries of what they can create, they often have a vision. For that, you need full creative control. You need a multiplayer canvas that’s performant and being in flow state, and you also need design and code and you need them to come together.

WSJLI: As the company scales, how is your approach to leadership changing?

Field: What’s top of mind for a while now is, how do you continue to get the benefits of the trust of small teams and the density of decision-making? And the speed of decision-making that you get when you have just a few people in a room at scale and how do you create the infrastructure to enable that? It’s so important to make the right decisions and that requires exploration, that requires design thinking... and really having that bird's-eye view, people in the organization who are closest to decision-making, they’re the ones who have the best context … if they have strong judgment. And so I try to reiterate to our team all the time that we need to hire people with great judgment and we need to evaluate people for judgment as well.

Too many agents? As more businesses adopt AI agents, some of their heaviest users are racing to prevent uncontrolled proliferation.

WSJ Leadership Institute's Belle Lin talked with a number of enterprises about how they're trying to tamp down on the problem, but without discouraging AI use

“Every day there’s quite literally new agents that are being created, and almost at every tier of the hierarchical structure,” said Mike Trkay, chief customer officer and CIO of Fair Isaac—which goes by the name FICO.

The company is now instituting governance practices to prevent too many AI agents from providing conflicting results for the same problem. “We recognize the risk that it presents,” Trkay said.

The issue, known as “AI agent sprawl,” stems partly from how easy it is for even nontechnical employees to create these independent AI bots, thanks to platforms like Anthropic’s Claude Cowork. 

At Lyft, Jason Vogrinec, the ride-hailing company’s executive vice president of AI transformation, said the company has rolled out Claude to its employees, and figured out an IT-approved way of sharing “skills,” the set of instructions that help Claude agents handle specific tasks.

Lyft also is working on creating a centralized platform with IT controls for all of its agents. Having too many agents creates a challenge for “a publicly traded organization with lots of regulatory obligations,” Vogrinec said.

But a recent conversation with Dan Diasio, EY Global Consulting AI Leader and Americas CTO at Ernst & Young, had me wondering if that’s wrong. Sometimes the people closest to the processes are actually the least qualified to truly disrupt them, he said.

This is the example Diasio gave me:

A team that was doing third party due diligence on vendors was asked to reimagine their workflow with AI. Their typical process was 67 steps and took about 45 hours of work. They came up with eight AI use cases that took their process from 45 to 32 hours.

Then a separate team, that for the most part knew nothing about the existing process, but included a couple experts from the original team, was given the same ask. That team took the process down to about 16 hours.

The key, Diasio said, is balancing deep expertise with, well, the opposite. These workflow shifts should be led by two people: “a deep domain expert and a person who has no idea what the function actually looks like but who’s really savvy with the tools.”

The challenge, he said, is making sure they’re on equal footing. “It’s hard to do because most of the time one is working for the other. It's creating that balance between the two where you get to much better answers of what the strategy is.”

— Isabelle Bousquette

• Security researchers used Anthropic's Mythos to help discover an exploit that bypasses Apple's advanced MacOS security protections, demonstrating the increasing capability of AI models to identify vulnerabilities. They plan to release details of their attack once Apple has patched the underlying issues.
• Thursday's IPO by AI chip company Cerebras Systems became the largest of 2026, raising $6.4 billion with shares surging nearly 70% on their first day of trading, signaling strong investor appetite for AI-focused companies, Barron's reports.
• A partnership between Apple and OpenAI may be on the rocks with the AI startup preparing potential legal action over breach of contract. Bloomberg reports that Apple's limited integration of ChatGPT failed to generate the expected billions in subscription revenue. Adding to the tension: Apple's plan to open its AI platforms to competing providers and OpenAI's recruitment of Apple engineers.
• A majority of Americans, some 70%, oppose building AI data centers in their local areas, with concerns centered on environmental impact, according to a March Gallup survey.