AI didn't raise creativity's ceiling—people pushing boldly did
Figma founder admits he was wrong when he predicted AI would raise the ceiling on creativity. AI only lowers the barrier to making things; only people bold enough to take real risks actually raise the ceiling.
The video won't play here. Listen to the audio instead:
The argument · tap a timestamp to hear it
Regulators killed a $20 billion deal by applying yesterday's market logic
Figma and Adobe signed a $20 billion acquisition agreement, but US, EU, and UK CMA regulators intervened on antitrust grounds, dragging out the review for 16 months. Dylan argues the regulators applied yesterday's market definitions to a market being redefined by AI right now. He cites the Roomba-Amazon case as a cautionary tale: FTC blocked it, Roomba eventually failed, and the market flowed to China—suggesting that antitrust intervention can harm the very competition it aims to protect.
— Dylan FieldLosing certainty gave 1,200 employees the one thing they needed most
The agreement collapsed on the first day of a company holiday. Dylan called an all-hands video meeting on Monday afternoon to announce the news. Employees' first reaction was not disappointment but relief—finally, they didn't have to live in uncertainty anymore. He then opened an exit door: if anyone had joined Figma only hoping to eventually work at Adobe, and wanted to leave, the company would pay them several months' salary with no hard feelings. The only condition was waiting six months before reapplying. In the end, only a small fraction chose to leave.
— Dylan FieldNew models disrupted design software; product moat won anyway
After Anthropic released Claude Design, the market immediately declared Figma dead. But Figma's next quarterly report showed 48 percent year-over-year revenue growth, bucking the prediction entirely. Dylan's explanation: the market conflated "model progress" with "product progress." Professional design tools like Figma have deep moats built on years of workflow refinement—not on underlying model capability. A new model-powered competitor struggles to replicate that accumulated expertise overnight.
— Dylan FieldCursor proved indie teams beat tech giants through sheer focus
Dylan cites Cursor as proof. By all rights it shouldn't have won—at the stage when Cursor lacked its own model, cutting-edge AI labs with in-house models were chasing the same lane. But Cursor obsessed over user experience and kept polishing product details, and it won anyway. His conclusion: SaaS isn't binary—"either everyone dies or everyone thrives." As long as you deeply understand your users and stay focused, small teams can still break through even with giants circling.
— Dylan FieldAI raised the floor of entry, not the ceiling of creation
Dylan admits he was wrong years ago when he said at Figma's Config conference that "AI will raise the ceiling of creativity." Now he sees it differently: AI only lowers the barrier to getting things made, letting more people produce more content. But the ceiling itself hasn't moved up by a millimeter. Only humans raise the ceiling—those willing to invest time, take real risks, and attempt something no one has ever done before.
— Dylan FieldLabeling AI-written work creates trust only for human-crafted contributions
Figma has a simple internal rule: anything written with AI must be labeled "written with AI" at the top of the document. Dylan says this started as a light joke, but people actually follow it. He sees it creating a trust hierarchy inside the organization—knowing that something was carefully hand-crafted by a person carries more weight, and that trust is the reward for doing things with your own hands.
— Dylan FieldAI excels where truth is verifiable; design remains a human domain
Dylan observes that models in verifiable, formal domains like mathematics and cybersecurity are stacking up capabilities at breakneck speed. But in creative and design domains—where you can't verify correctness—progress has been merely incremental. This might actually reveal humanity's remaining role in the AI toolkit: people brainstorm and design; AI executes more of the work. Design and creativity remain fundamentally human acts.
— Dylan FieldSolving 700 math problems solves away something precious and irreplaceable
The night before recording, OpenAI released over 700 papers cracking famous unsolved math problems. Dylan called it "perhaps the greatest day of discovery in human history," but admitted a certain melancholy underneath—as if humanity gained something and lost something at once. The mystery and wonder of these mathematical puzzles is being stripped away and solved. There's triumph in it, and a tinge of loss.
— Dylan FieldIn their own words · checked verbatim
this was when everyone had the Saspocalypse thesis in general. And I was a big subscriber to Saspocalypse.
Dylan Field14:16
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 but yet cursor really obsessed over the user
Dylan Field17:18
The ceiling has stayed the same. AI is not raising the ceiling. Humans are raising the ceiling, though.
Dylan Field20:30
So my view is design is the new code. It's the layer above.
Dylan Field21:32
hey, if you are using AI in your writing of any kind, just put at the top written with AI.
Dylan Field24:33
models keep getting so good when it comes to these formal domains. It's math, cybersecurity. Like, the IQ points just stack and stack and stack. And then for creativity and for design, it's very incremental progress overall.
Dylan Field32:43
I thought it was the greatest day of discovery in human history.
Dylan Field38:06
Figures
| Adobe's acquisition offer for Figma | $20 billion | 3:03 |
| Regulatory review duration | 16 months | 8:06 |
| Figma headcount when deal fell through | approximately 1,200 | 6:05 |
| Figma revenue growth after Claude Design launch | 48% | 14:16 |
| Figma founded | August 2012 | 2:00 |
| OpenAI math papers on breakthroughs | 700+ | 38:06 |
Glossary
- SaaS apocalypse
- The thesis that AI will render traditional subscription software entirely obsolete, replaced by AI-native applications
- Overton window
- The range of ideas a society considers acceptable for public discussion in a given moment; ideas outside it are treated as taboo
- MCP
- Model Context Protocol—a standard that lets AI models connect to external tools and data sources; Figma used it to integrate with Anthropic
How to listen
Founders building design tools or enterprise software, investors, and product managers trying to understand design's value in the AI era.
Opening pleasantries and sponsor messages (around 0:00–2:00) can be skipped.