Dylan Field: Is AI Replacing Designers?, Design is the New Code & Democratizing Creativity

Anthropic

Across the shows: 22 speakers in other episodes, 15 bullish and 35 bearish calls, past 30 days

• Dylan Field · FigmaNeutral

  Neutral on Claude Design; its broad scope may win some areas, but professional design tools are hard to build.

  Field says Anthropic's Claude Design has been built as a broad product that tries to do many things, including posters and graphic design scenarios. He thinks it may do well in some areas and not in others. He says he is 'not bullish or bearish' on the product. His reasoning is that professional design tools are hard to make because there are many design categories, and the feel and mechanics of a tool must be right for different users and tasks. Lab-built tools can therefore be strong in parts of the market without being broadly dominant.

  “I'm not bullish or bearish on the product. I think it's just like hard to make good design tools professionals. Yeah. There's a lot of things to design.”at 16:15

Anysphere Inc. (Cursor)

• Dylan Field · FigmaBullish

  Cursor beat frontier labs with their own models by obsessing over users, showing specialised AI tools can win.

  Field cites Cursor as the best example of a company that should not have grown as it did, because frontier labs were chasing it with their own models before Cursor had a model of its own. He argues Cursor succeeded because it focused on the user, identified the gaps in their workflow, and made sure it was doing the right thing for customers. He says this shows that specialised, user-obsessed AI software can keep winning against model makers, and he expects the pattern to repeat across many company types.

  “Cursor, like, by all means, should not have grown the way they did. You've got all these frontier labs that are chasing them with their own models before Cursor had a model of any kind.”at 17:31

OpenAI

Across the shows: 15 speakers in other episodes, 5 bullish and 14 bearish calls, past 30 days

• Dylan Field · FigmaBullish

  OpenAI's release of 700+ math papers on famous problems is a major demonstration of AI capability.

  Field cites OpenAI's release of over 700 papers covering very famous unsolved math problems as the event that hit home most for him. He says it was 'a show of something' and calls it the greatest day of discovery in human history, which he calls factually accurate. He pairs the excitement with a sense that something was lost, so the view is enthusiastic about the capability but mixed on its broader effects. No market or financial implication is stated.

  “nothing hit home for me more than the math drop last night from OpenAI of 700 plus papers on all sorts of very famous problems. That that was a show of something.”at 38:32

Software & SaaS

• David Friedberg · The Production Board

  Founder-led SaaS with embedded user experience can be distinguished from the SaaS-apocalypse group.

  Friedberg says he was a big subscriber to the SaaS-apocalypse thesis at the time Figma's quarterly results came out, showing 48% growth and accelerating business momentum. He argues that SaaS companies can be separated along lines such as founder leadership and embedded user experience. He adds that founders who know how to use AI to accelerate and reinvent their business are particularly well placed. He uses Figma's growth to show that the category is not uniformly doomed.

  “I do think that there's a bunch of ways to distinguish SaaS companies, one of which I think is founder led. And I think that there's user experience that matters a lot, embedded user experience”at 14:34

• Dylan Field · Figma

  SaaS is not dead; not all SaaS firms are fine, but the software category is not binary.

  Field rejects the 'SaaS apocalypse' thesis as too binary. He says there are more ways for software companies to sustain themselves than that thesis implies. Specialisation on a target audience and deep knowledge of users matter. As evidence, he points to the exponential growth in App Store listings, while overall app usage stays flat. This means more apps compete for the same users, so differentiation is needed. He also says not all SaaS companies are fine, so the sector is mixed rather than uniformly threatened or uniformly safe.

  “SaaS is not dead. ...it's not true that all SaaS companies are fine. Yeah. And it's also not true that all SaaS companies are dead.”at 18:03

Antitrust & tech regulation

• Dylan Field · Figma

  Regulators' Adobe-Figma blocking arguments were poor, and sixteen-month reviews are punishing and need fixing.

  Field says he doesn't believe the Adobe deal was worth blocking on antitrust grounds, and that regulators' arguments were poor and focused on past battles rather than the future. He says the sixteen-month regulatory review was punishing for any company, and the process needs fixing, especially if a deal is not actually anticompetitive. He cites Roomba as an example: a small deal that was blocked (the Amazon acquisition of iRobot), after which the company went bankrupt and the market went to China. He argues this outcome made no sense from a US standpoint, and that regulators define markets looking backwards rather than forwards.

  “I still don't believe the Adobe thing was for what's worth... but I thought their arguments were poor. And they were fighting past battles, not looking at the future and what this all would become.”at 7:31

AI & jobs

• Dylan Field · Figma

  AI lowers the creative floor, but humans raise the ceiling, so design becomes the differentiator.

  Field argues that AI makes production of code, motion, video and design easier and more democratised, which lowers the floor. But he says he was wrong to think AI raises the ceiling, since the ceiling has stayed the same. Humans raise it when they invest time, are bold and try things no one has done before. With code cheap, the scarce layer is design, point of view and brand, which he calls 'the layer above.' He points to the exponential growth of App Store listings, with flat app usage, as evidence that differentiation is now the key need.

  “we're lowering the floor with AI... Everyone can be part of that, and it's more democratized. ...AI is not raising the ceiling. Humans are raising the ceiling, though.”at 20:02

AI: overall outlook

• Dylan Field · Figma

  AI gains compound in verifiable domains like math and cyber but stay incremental in creative design.

  Field observes that models keep improving quickly in formal domains such as math and cybersecurity, where 'IQ points just stack.' In creativity and design, progress has been very incremental. He thinks this is because creative and design domains are nonverifiable, so there is no clear reward signal to optimise against. He suggests this could change at some point, but for now the gap is large, and the 'novelty' users feel around each new model drop fades within days.

  “IQ points just stack and stack and stack. And then for creativity and for design, it's very incremental progress overall. ...I think the fact that these are nonverifiable domains has something to do with it.”at 33:25