I’m talking to the people building what’s next.

Thaddeus and I go way back. This was one of those calls where you catch up for five minutes and then accidentally say something that sticks for a week. He was telling me about his product — Dobbin builds AI strategy tools — and the conversation drifted to how people actually evaluate what AI tells them. I said what I've been thinking for a while: for the last 15, 20 years, we've known how to look at Google results and instantly sort signal from noise. Which link is an ad, which domain is credible, which result is SEO gaming. That skill is invisible because it's so practiced. But with AI? There's no equivalent filter. No muscle memory. No pattern to match against. The confidence that makes AI feel useful is the same thing that makes it impossible to evaluate. The test I keep coming back to: I don't want to ask AI how to play basketball. I want to ask Michael Jordan. The distinction matters. "AI" is a voice without a framework. Michael Jordan is a framework you can evaluate — his experience, his era, his biases, his blind spots. You can learn from him and disagree with him at the same time. You can't disagree with a confident paragraph that has no author. That's why the Symposium shows its reasoning. When Taleb pushes back on a question, you can click through to his axiom. The AI isn't pretending to be anyone — it's modeling published frameworks, and it shows you the model. That gives you something to be literate about.

Jose and I haven't talked in years. We caught up the way you do when enough time has passed that you skip the highlight reel and just tell the truth. His startup has about a month of runway. I left my last job six months ago and I'm watching savings drain. Neither of us has it figured out. Both of us are old enough that we're supposed to. Everyone in his circle tells him the same thing: dude, you always hit the last shot. Ten seconds on the clock, you pull it off. He's tired of it. So am I. There's a specific exhaustion that comes from being the person who always lands on their feet — because landing on your feet is what you do instead of building a floor. I told him something I haven't said to many people. For years, I manufactured this instruction manual — the Ian manual. Printed it, handed it out. This is how you deal with me: I'm good, I got it, don't worry about it. And everyone followed it. Page 62: don't worry about Ian. Then you feel empty and pissed at people for doing exactly what you told them to do. I've been going back to people with the new manual. That part has been really interesting. Jose said if his startup doesn't make it, he wants to do the postmortem differently. Not a LinkedIn "sad to announce" post. A proper journalistic piece — interview the team, the investors who backed them, the ones who didn't. Tell the real story of what it looks like to have an idea, build something real, get people using it, and have it still not work out. I've never heard anyone describe a failure narrative with that much intention. Most people eulogize their companies. He wants to autopsy his.

Zapier's internal framework names four ingredients for AI transformation: leadership, talent, culture, tools and governance. Two of four are human. Every panelist — NerdWallet's CPO, DoorDash's head of AI transformation, Webflow's chief AI evangelist — independently described the same pattern. The technology part is the solved problem. The hard part is getting 350 marketers to stop being afraid, or getting decentralized business units to share what they're learning, or convincing leadership that "hours saved" is not the metric. DoorDash's AI leadership model: clear direction, clear governance, clear accountability, but distributed ownership. Each department owns its own priorities. NerdWallet mirrors its org design — portfolio-style business units with their own P&Ls run their own transformation. Webflow's chief AI evangelist is a made-up title born from previous experience, not an org chart. Three different structures, same principle: the people closest to the work know where the biggest opportunities are. Whatever your leadership model looks like, it should reflect that. The Zapier host said something that landed: at the turn of 2026, they stopped counting hours saved as a value metric. Hours saved is fuel for potential impact — but only if it converts to customer success, revenue, or product quality. By itself, it's just a number people put on slides. DoorDash is stepping through the same realization. First you save hours. Then you ask: what did those hours become?

I can't listen to words while I work. Neither can Thos. I put on Sufi music — keeps me in a communication vibe without engaging the language part of my brain. Thos found his version during the pandemic: a German Peloton instructor. Couldn't understand a word, but you get the intent. Look at the numbers, figure out what to do. All the motivation without the insufferable American positivity. I built something 20 years ago — a recommendation engine that mapped work types to music. Spreadsheets with urgency? Russian Circles. Coding? Vow — thunderous, screamy, something that occupies the part of your brain that would otherwise wander. Audio weighted blanket — that's the phrase we landed on, and I haven't heard a better description of what music does for focused work. It's not ambiance. It's cognitive infrastructure. You're medicating one part of your brain into cooperation so another part can lock in. Everyone has their version of this but nobody talks about it because it sounds unserious. It's not. The conditions you construct around your focus matter as much as what you're focusing on. And the people who know theirs have an advantage nobody teaches.

Robert wrote the book on this. Literally. His argument isn't that compliance is bad — it's that it's not enough. The box-checking version of privacy lets you do shitty things. You can satisfy GDPR, pass a security audit, and still build something that exploits users, because the regulations don't ask whether your practices are actually good. They ask whether you've technically satisfied a minimum bar. Companies spend enormous energy finding the slippery route that technically meets the requirement while violating the intent. He draws the same parallel I always draw with accessibility: most design orgs treat both as the audit at 11:59. Someone says, "can you take a quick look at this from a privacy perspective?" after the product is functionally done. You're looking at a stack of decisions you can't unwind. Robert's been in rooms where someone hands him 15 form fields and asks which ones are risky. The real question isn't which fields are risky in isolation — it's why do you need any of them. Social security number is obvious. But asking for someone's state of residence or gender can be equally loaded depending on who's filling out the form and what their life looks like right now.

Eric saw a musician at Eddie's Attic last weekend — a guy named Mike Viola who's been behind some of the biggest artists you've never heard of. Eric assumed the guy was in his late 40s. Looked him up after the show. He's 59. Eric stayed up until one in the morning writing because something cracked open. We both grew up in a time where our jobs defined every aspect of us. That's what we were taught — know what you want to be, pursue it, let it consume you. Somewhere in the last decade, both of us lost the creative outlets that made us interesting in the first place. Eric stopped writing music. I stopped painting. Not because we chose to, but because careers filled every gap until there was no room left for the thing that wasn't productive. Eric had a designer on his team at Salesforce who retired recently — a lead, never tried to climb the ladder. Just showed up, was clear about what he'd deliver, executed, and went home to walk his dog. No ambition theater. No performance. Eric said that guy taught him more about boundaries than any manager he ever had.

Vas and I go back to my agency days in New York. We'd sit for hours — not working through client problems, just working through problems. Philosophy, ambition, fear, what we were building and why. The kind of conversation where you say something half-formed and the other person helps you finish it instead of judging it. That's harder to find now. Careers make you guarded. Success teaches you to perform instead of explore. The conversations that used to happen naturally — at a bar, after a pitch, on a long cab ride — now require real intention. Most people stop making the space. The Salon exists because I think we have to.

Laura's been in the design industry long enough to have lived through cycles — the move from craft to digital, from agency to in-house, from bespoke to systematic. She sees another one forming. The current AI-everything push is creating a sameness that consumers are already reacting to. Gucci just got dragged for using AI-generated images in a campaign for a house built on craftsmanship. People can feel the absence of a hand. Her argument isn't anti-AI. It's pendulum physics. Every era of mechanization has produced a counter-movement toward the handmade, the imperfect, the identifiably human. Arts and Crafts followed industrialization. Vinyl followed streaming. The more everything looks generated, the more a human fingerprint becomes the luxury signal. Laura thinks we're 18 months from "made by humans" being a brand differentiator the way "organic" became one for food.