I went in expecting a room full of designers quietly panicking about AI. That is the story everywhere else. Every feed I read is some version of “is this the end of the craft,” so I showed up braced for a wake.

I got the opposite. The people on that stage were not scared of AI. They were excited by it, and most of them were already busy making things with it. This is the longer version of what I saw, talk by talk, for anyone who wants more than the highlight reel.

AI as a door that opens

The clearest argument for optimism came from Martin LeBlanc. He always wanted to be an animator and never had the talent to land somewhere like Pixar, so he let the dream go. AI gave it back. He walked through how he now builds short films a few seconds at a time, generating a beat, feeding that clip back in as a reference, and telling the tool to continue the story until he has a whole scene. He uses world models to keep a room spatially consistent when the camera moves, and he brings his own hand into it by drawing characters on his iPad and turning those sketches into a full cast. The point that landed was simple. He could not have made any of this before. The tools did not flatten his craft. They opened a door that had been closed his entire career.

That theme showed up again in Phil Hedayatnia’s talk. Phil runs a design team of 40 across 14 countries and builds a research tool called Meld. He told a story about visiting ByteDance and learning they moved to designing directly in code five years ago, because Figma never launched in China. What surprised him was the result. They did not hire fewer designers. They gave designers more responsibility and pointed them at understanding the psychology of the people they were building for. His read on the future is that abundant intelligence in the process does not shrink the design role, it moves it up the stack.

We have been here before

Brian Benitez zoomed out and called this a designer’s playground. His evidence was the early iPhone. When the App Store opened, the interaction patterns we now treat as obvious were not in any guideline. Pull to refresh was invented by a designer named Loren Brichter. The hamburger menu came from Facebook needing to fit too much into one screen. Designers made those up by playing with a new device, and the people experimenting early wrote the defaults everyone copied later. Brian thinks AI is the same kind of open space, and the same opportunity.

Pauline Wee took on the fear directly. She talked about declinism, the feeling that things are always getting worse and it is already over. Then she traced how computer vision art went from groundbreaking in 1999, to boring within a few years as everyone copied it, to cool again once AI arrived, to lame again once social media saturated it. Her conclusion was that it is never over, the cycle just keeps turning. She also asked a question I have not stopped thinking about: what does your body of work look like across a whole life, not this week? The unit that matters is the collection, not the individual post.

Why the human still matters

The optimism was not naive. Underneath it, almost every speaker made a version of the same claim: the tools are exciting because they still need a person with taste, judgment, and a life.

Wes Bancroft made the case with the most heart. His talk was that work itself is an act of love, if you let it be. His central idea is that every time you make something, there are two outputs. There is the thing that leaves your hands, and there is the person you become by making it. He argued we have spent years optimizing the first output and forgetting the second. AI does not change that math. If it takes over more of the first output, the second one becomes the whole point. He tied it off with the idea that the work was never really about making things, it was also about making us.

Flora Guo put the same idea in sharper language. She is a design engineer, and she was blunt that prompting something into existence does not make you one. Her strongest line was that the model has hidden layers but no body. Cognition is not only in the head, it is distributed through the body, and an LLM has no perception, no context, no life to pull from. That is why raw generated work reads as slop. She wants tools that work with you rather than for you, and she ended on the reminder to go outside and touch grass, meaning your material comes from living, not from the canvas.

Dan Hollick made the technical version of the argument, and it was one of the best talks of the day. He spent most of his time showing how much sits underneath the tools we use without thinking: the pen tool is Bezier curves, fonts are collections of those curves, resizing an image is an interpolation problem, blur is kernel convolution, JPEG and GIF are just clever ways of throwing away data your eye will not miss. Then he flipped it. With AI, the new medium of production is code and the new abstraction is language. His conclusion was counterintuitive and convincing. We are going to get more technical, not less, because language is a lossy way to describe what you want, and the more precise your understanding, the better your results. Knowing how things work is what separates good output from garbage.

Phil closed his talk on the most practical piece of advice at the conference. The best thing a designer can do right now is spend as little time as possible in front of a laptop, and as much as possible in the real world, meeting the people the work is for. He is not anti-tool. He built one. He just thinks the scarce input now is context, and context comes from being somewhere.

Michael Riddering told the story of chasing the design engineer title, finally reaching it with AI, and finding it empty. He could build the thing now, but building was only the surface. His argument was that a designer’s real value sits underneath the craft, in understanding what you are building, using research and judgment to know which bets are the big ones, and using your influence to put that value into the organization. The craft is fun. Knowing what is worth making, and why, is more fun, and it is the part AI cannot reach yet. Designers are the ones positioned to find that value and express it.

The best talk had no AI in it

David Stoddard did a fireside interview about building Brixton, the apparel brand he started in a garage in Oceanside in 2004 and grew to around 120 people and four offices before selling in 2014. There were no models, no agents, no prompts. There was a guy who refused to run the same ads as everyone else, who used black and white portraits when the whole industry used color action shots, and who used a catalog four times a year to tell the story of a brand that spoke to skaters, surfers, and punks at once. He was honest that luck and timing mattered, and his practical version of that was to put yourself in the environment where your thing can grow, then build relationships and ask.

I think it was the most useful talk of the day precisely because it never mentioned the thing everyone else was circling. It was a reminder that taste, authenticity, and knowing who you are do not change when the tools do. Those are the parts of the job that were always the job.

What the panels argued about

The two panels pulled together founders and leaders from companies like Instagram, Magic Patterns, Pen.dev, and Zoah, along with independent studio owners, brand designers, and other people who think about this for a living. Between them they filled in the harder, less tidy questions.

The design leaders panel spent its time on where the roles are going. One argument was that design splits into two archetypes, the product builder who thinks in strategy and ships fast, and the taste maker whose craft is strong enough that AI cannot replicate it. There was a real disagreement about which function survives longest, product or design, and a healthy dose of skepticism about the agent future. The sharpest version of that skepticism: most people do not want their bank balance or their house behind a chat box. We want stable, predictable interfaces for most things, and chat is only good for the cases where you roughly know what you want but not exactly.

The panel Q&A got tense in a good way. Someone in the audience called the leaders out for sounding like they were closing the gate behind them on junior designers. The panel took it seriously. They admitted design hiring is broken, that big companies have cut junior and mid roles and only want senior people, and that this predates AI. The most useful distinction that came out of it was mentorship versus sponsorship. Mentorship is showing someone what you do. Sponsorship is betting on them, opening doors, making introductions. The industry has plenty of the first and not enough of the second, and one leader put it bluntly: if you cannot train a junior, are you actually good? They also landed on why humans stay in the loop, and it was not only about judgment. It was about liability. You cannot put an agent in jail. We have centuries of courts for human decisions and nothing equivalent for machines, so a person stays responsible.

The tool makers panel dug into the tension every design tool is trying to solve: exploration versus production. The old complaint is that a designer spends hours making a mockup, and then someone has to rebuild it in code because code is the real source of truth. The debate was whether a code based canvas fixes that, or whether it kills the freedom of a blank canvas. Constraints help code and hurt exploration, and the craft is knowing when you need which. Their Q&A got into the business side, including the shift to usage based pricing and the discomfort of building your product on top of AI you do not fully understand or own. The reframe there was that founders have always built on things outside their control, a supplier, an engineer who might quit, but the caveat stands that you cannot fully trust an AI built foundation yet.

What I left with

I have not felt this optimistic about design in a while. The through line across every talk was that the tools are opening doors that used to be locked, for people like Martin who were told they would never get in, and that walking through those doors still takes a person with taste and something to say. The homework the conference kept handing out was to go make something, and to live a life outside the screen worth making things about.

Thanks to Drew Wilson for putting it on. Here’s to seeing everyone next year!