Dylan Field: Design Is the New Code

The interview happens on a patio in Yountville, California, a few drinks into the evening, after Dylan Field has already spent the afternoon on another stage talking about the same subject he is about to talk about again. Field runs Figma, the design software company that twelve years ago started as a bet that browsers could handle real creative work. He is relaxed enough to joke about being "three drinks in," but the idea he keeps returning to is not loose at all. It is a clean, almost provocative claim: design is the new code.

What he means is specific. As AI coding tools from companies like OpenAI and Anthropic get better at writing software, the nature of programming is changing, and the value tied to differentiation rises toward the top (01:10). That does not mean engineers disappear, Field is careful to say Figma still needs them. But the layer where value concentrates moves upward, from the mechanics of building something to the feel of it: the brand, the boldness, the emotional reaction a product provokes (01:34). If anyone can generate a working app, the thing that separates a product people love from one they ignore is no longer whether it works. It is whether someone had a point of view.

Figma itself is Field's working definition of that idea. Founded in August 2012, it is not just a tool for professional designers, people use it for slides, whiteboarding, diagramming, almost anything visual (01:52). Field's own gloss on the word "design" is disarmingly simple: "creativity meets problem-solving" (03:03). That phrase matters later, because it is the hinge on which his whole argument about AI turns: creativity and problem-solving are not the same skill, and AI is reshaping only one of them.

A Company Learns to Wait

Before any of that, though, there is a story about waiting. A few years back, Adobe offered to buy Figma for $20 billion, a deal large enough to be one of the biggest acquisitions in software history (03:20). Figma said yes. Then, in Field's words, "various world governments decided that they didn't like it." Regulators in the United States, the European Union, and the UK's Competition and Markets Authority opened reviews. Antitrust law, the set of rules meant to stop mergers that reduce competition, is built for a narrower, slower-moving idea of markets than the one Field says he was actually operating in. His argument, half business logic and half grievance, is that regulators defined the market by looking backward at design software history instead of forward at a market that was really "AI," vast and still forming (09:05).

The review dragged on for sixteen months before the deal was blocked (08:24). Field does not hide how brutal that stretch was for a company of roughly 1,100 to 1,200 employees who had been told they were about to join Adobe.

"16 months of regulatory review for any company is punishing." — Dylan Field [08:24]

He reaches for an odd but pointed comparison: the robot vacuum maker Roomba, whose acquisition by Amazon regulators also blocked, supposedly to protect competition, after which Roomba went bankrupt and the market went largely to Chinese manufacturers (08:24, 08:45). His point is not subtle: idealism aimed at a specific company can hand the whole category to someone else entirely.

The all-hands meeting where Field told Figma's staff the deal was dead happened, almost comically, on the first day of the company's winter break (09:53). What struck him was not disappointment but relief, not relief at avoiding Adobe specifically, but relief at the ambiguity finally ending (10:17). He then made an offer: anyone who had joined expecting to work at Adobe, or who was simply tired, could leave with several months of pay and no hard feelings, with one condition, a six-month wait before they could reapply for a job at Figma (11:18). Few people took it. The company kept building, went public, and later posted 48 percent year-over-year growth, an acceleration that came, notably, right after Anthropic's Claude Design product launched and online commentary briefly declared Figma finished (14:07).

The Floor and the Ceiling

That episode sets up Field's larger argument about what AI is actually doing to creative work. A few years ago, at Figma's own user conference, he predicted that AI would raise the ceiling of what people could make. He now says he was wrong.

"The ceiling has stayed the same. AI is not raising the ceiling. Humans are raising the ceiling now." — Dylan Field [20:41]

AI, in his telling, is lowering the floor: it makes basic production of code, video, or design available to anyone, which is genuinely democratizing (20:02). But the floor is not the same as the top. The App Store keeps producing exponentially more software, yet how much people actually use apps has stayed roughly flat (21:32), which tells Field that abundance of production is not the same as abundance of value. The scarce thing, now that everyone can hit an average result, is a human willing to be bold, to have a point of view that is not the consensus one (22:28). He links this to a broader anxiety about AI-saturated communication: a flood of AI-written decks and messages where, as he jokes, "I don't even know if the person who made it read it" (24:40). Figma's response was a small, literal policy: any internal writing that used AI gets labeled "Written with AI" at the top, so colleagues know how much human attention actually went into it (25:14).

Underneath this is a claim about what makes human creativity different from a model's output at all. Field describes human reward and loss functions, the internal signals that make us repeat or avoid behavior, as fundamentally different from how current AI systems learn: a person splashed with water once will avoid repeating the mistake, where a model trained with reinforcement learning needs the correction many times over (29:48). He connects this to novelty-seeking, the idea that people are drawn to the genuinely new rather than the statistically likely, while a model's outputs are shaped by the distribution of its training data, by definition closer to the average of what has already been made (28:50). Design, in his view, asks for something harder than either alone: holding a user's needs, a brand, a business problem, and an aesthetic judgment all at once, a kind of combinatorial complexity he finds "fascinating" rather than threatening (30:42).

Near the end, Field frames all of this as a choice that predates AI and will outlast any particular model.

"As a human, in general, pre-AI, post-AI, you have a choice. Do you wanna consume or create?" — Dylan Field [36:33]

He is unsentimental about what easy AI use looks like: letting a model think for you, taking the path of least resistance. The harder path, using AI to learn things that would otherwise take too long to master, is the one he says is worth taking. The conversation closes on an unresolved note that feels true to the subject. Field describes OpenAI's recent release of solutions to more than 700 papers' worth of famous unsolved math problems, a single overnight event he calls, half in awe, "the greatest day of discovery in human history" (38:28). But he admits the feeling was mixed: exciting and, in some way he cannot quite name, sad, as if a bit of mystery had been spent. That tension, between a floor that keeps dropping and a ceiling that only a person can raise, is the real subject of the hour, long after the Adobe deal and the $20 billion number have been filed away as history.