Design’s staying power in an AI-first era
For Dylan Field, Figma’s co-founder and CEO, the question isn’t whether AI tools will displace designers, but whether rapid iteration will come at the cost of craft. In a conversation with Garry Tan, president and CEO of Y Combinator, Field made the case that the current wave of artificial intelligence tools—however capable—still belongs firmly in the “tool category.” They augment specific tasks and expand what designers can attempt, but they have not yet absorbed the contextual judgment that makes for finished work. Field’s mental model is the “idea maze”: branching paths of ideation where AI grants more breadth, but depth still takes effort.
The gap between prototyping and finishing
Field is blunt about the current disconnect in AI-assisted work. The language creeping into build culture—“vibe coding,” “getting locked in,” “I’m cooking”—reflects a genuine pleasure in fast feedback loops, but the tools often shine brightest early in the process. The harder problem is taking an idea past the prototype and getting it to a finish line. That gap, according to Field, is not unique to design; it shows up in code generation too.
The playfulness that propels a product forward early can become a liability later. Everyone can mock up novel ideas now, but developing a first draft into something polished and usable is where craft still matters. Field noted there is a productive rhythm between the speed of exploration and slower, deeper rounds of refinement.
The differentiation argument for design
As the cost of building software drops, the incentive to invest in design should increase. Field believes an "age of AI" presents an opportunity for design to become a broader competitive differentiator. He emphasized that the question isn't can you make something work, but how it works that separates good products from forgettable ones.
Yet he pushes back on any sentiment that designers are becoming obsolete. In fact, he said his conviction in the value of the designer’s role has grown over time. Field said he expects the title itself to shift meaning: developers already working with generative tools may describe themselves as designers more often, not because they stopped writing code but because increasingly they make holistic decisions about the product experience.
Field also challenged the common assumption that text-based prompting is the endpoint for AI interfaces. Drawing an analogy to the days of telnet, he views current prompting mechanics as an early phase of a spectrum. Predicting a range of interaction models, including renewed attention on visual design, AR, VR, and other "dynamic" surfaces, he said more than a couple of categories and styles will likely coexist significantly in the near future. New patterns will expand "the possibility space" for how people work.
On the limits of today’s models, Field speculated about why they stumble when asked to imagine polished output. Current AI modeling approaches remain polarized—diffusion-based on the art side and LLM-based on problem-solving—with little evidence their integration has been cracked. That’s not a trivial issue because Field sees design as problem-solving informed by art. A designer’s job is converging cultural cues, brand awareness, and context into a coherent read. The context a trained eye brings exists far outside the two-line prompt box.
Figma’s winding start
Field offered a trip through Figma’s messy foundations. Late in 2011, while still an undergraduate at Brown, he and original co-founder Evan Wallace (then his teaching assistant) began asking what open, shifting technologies—specifically WebGL—enabled. Their early explorations ran a wide and sometimes goofy gamut: the smallest idea that will not die in Figma lore being a meme generator, which Field characterized as “the low point.” They built an impressive web photo editor too, but one that lacked sufficient reason to exist as a browser-based product.
When Adobe announced its sunset of Fireworks, the unrelated story clicked for Field: if creating software got easier, good design should grow in economic value. What felt like a niche market at the time turned out to be one that scaled quickly.
Convincing investors proved hard. Field admitted their seed pitch in June 2013 was “all over the place,” with much of the narrative built around a projected market expansion. Their saving grace: early customers pulled Figma into gradual improvement. First adopters were often minimalist companies (names include Notion and Coda) looking for cloud-first performance. Field relayed an early anecdote about leaving a first customer’s office and turning back immediately after receiving a bug report about broken fonts—details matter when retaining nascent trust.
The tool’s December 2015 beta lacked the feature set that eventually define it: multiplayer was absent. The real breakthrough, a collaboration engine built with Wallace, didn’t appear until the GA launch in October 2016. Reactions were mixed — divisive commentary like “a camel is a horse designed by committee” rolled in from the design community. Field read this pushback as reflecting a more traditional “agency mindset” where designers perform in the corner and reveal a finished project, rather than the collaborative product team method Figma was pushing. In a farcical twist befitting startup lore, early users mocking the product by holding a “design party” in a shared file offered unexpected proof of concept. Keeping that one massive file alive for 48 hours forced some firefighting, though the occasion showcased the tool’s promise far better than any pitch deck could. That live jamming, and the accidental community energy that surfaced, did the talking for Figma, ultimately driving both adoption and the direction of its core product philosophy.
Design culture and the maker mindset
Dylan F.
Figma has benefited from a strong bench of designers, and the ongoing challenge is balancing approachability with power — making the simple thing simple while keeping the complex thing possible. That tension tends to resolve best when the best people are involved, and when the whole company is invited to contribute. Internal initiatives like Maker Week have produced some of Figma’s most successful features, reinforcing a culture where everyone has a hand in improving the product.
Scaling beyond the founder’s loop
Dylan F.
Founders who have made it from zero to one face a different problem at scale: recognizing the loop they are stuck in. Be self-aware about what you are spending the most time on and then replace yourself in that task. If you have the resources, do it directly; if not, get them through profitability, fundraising, or creative resourcefulness. Repeatedly applying that loop builds the team and establishes proper delegation. The bigger risk is becoming so reactive that the organization never improves itself.
Speed over long build cycles
Dylan F.
The calculus for new founders has shifted with AI. A two-year build cycle is rarely necessary anymore. Many founders still propose one, and having done it myself, my advice is to move faster. If there is a way to compress the timeline, take it — the advantage goes to those who launch sooner rather than later.
