Re-founder

Sometime around 2019, Brian Chesky had a dream. Airbnb had roughly 7,000 employees by then, and Chesky, a RISD-trained industrial designer, felt like he was driving a car with no steering wheel: he would say turn left, and the company would go right. In the dream he left for ten years and came back to a place he no longer recognized, a giant political bureaucracy, and then the dream delivered its punchline. The person who had built the bureaucracy was him. "It was me the whole time."¹

A few months later the pandemic took 80 percent of the business in eight weeks, and the rumination stopped. What Chesky did next is the whole archetype. He did not batten down and delegate his way through the storm. He did the opposite: stopped delegating, climbed back into the details, and reviewed essentially everything at Airbnb for the next two to three years at 100 hours a week. "Before I empower people, I need to know what's going on."¹

This is the re-founder. Not the coach who learns that founder presence subtracts value and steps back from the work. The re-founder is the founder who, having built something large, notices that scale quietly pulls the startup out of a company the way gravity pulls everything toward the floor, and decides to put it back by hand, on purpose, again and again. Chesky compresses the doctrine into six words. "Great leadership is presence, not absence."²

The plate carries two anchors because the pattern has two settings. Chesky reinstalls the pace continuously, by standing in the room. Tobias Lutke ran the same correction once, at the same moment in the same year, and then welded it down so it could not come loose again.

The startup leaks out by default. He puts it back by hand.

Scale is gravity, and the pull is the enemy. The re-founder's founding belief is that a large company does not lose its edge through bad luck or bad hires. It loses it by default, continuously, the way anything loses altitude. Airbnb's growth had sagged to 10 percent against nearly $100B in annual gross bookings, and Chesky's read was that a marketplace at that size is "like gravity," almost impossible to re-accelerate once it starts coming down.² Most CEOs file that under physics and manage the decline. The re-founder files it under enemy.

Renovate one room, never the whole house. The method is deliberately unglamorous. No company-wide transformation, no consultants, no sweeping reorg. "Don't renovate the whole house. Pick one room and make it perfect, and then go room to room."² At Airbnb the first room was Project Hawaii: ten or twelve people on a single surface, the guest conversion rate, run "as if we're at Rouse Street," the apartment where two founders once inflated air mattresses to make rent. The room paid, by Chesky's own math, the equivalent of $200 million in revenue its first year and $400 or $500 million its second,¹ and the whole company's growth came back from 10 percent to 18 on a lag.² Small team, narrow surface, founder back in the room.

Teach the pace until it is muscle memory, then move on. The reviews ran weekly, sometimes daily, for as long as it took. He handpicked the team from people already there rather than replacing anyone, and he held the bar in person until a new standard became reflex. Some found it uncomfortable and left. Most stayed and found the work easier, because a CEO in the room means less bureaucracy, not more. Permanent supervision was never the point: once the pace runs on "muscle memory,"² he steps back and carries it into the next room.

Presence is the job, and soon nobody gets to manage only people. The doctrine has an edge. A generation of management advice told CEOs to trust their people and get out of the way, and Chesky's entire position is that trust and absence are not the same thing. "If you're a cavalry general, you should be on a horse on the battlefield."² He pushes it into the AI era, where "there should be no pure people managers,"² because everyone is now close enough to the data and the agents to be in the work directly (his evidence: 60 percent of Airbnb's code is now written by AI, twice his benchmark of competitors and peers²).

The pace is the product.

He kept the wartime posture on purpose

The re-founder is a chosen archetype, and the choice only makes sense against what Chesky was choosing to stop being. By his own account he was a bad CEO for most of the 2010s. "No one is born a good CEO. The job is completely counterintuitive and almost all of your intuition is wrong."¹ He had run Airbnb by learning as he went, which is exactly right for a founder and nearly fatal for a CEO, because a founder's mistake costs a weekend and a CEO's mistake, a wrong executive who quietly builds an empire and leaves, costs four years to unwind, his own number.

The pandemic ended the drift, but the pandemic is not the archetype. Plenty of founders went to war in 2020 and went back to delegating the moment it was safe. Chesky did not. He decided the wartime way of running the company was also the peacetime way, and he kept re-founding room by room even when nothing was on fire. The intellectual permission arrived secondhand. Around the time of the pandemic he hired Hiroki Asai, Steve Jobs' creative director, a man so far from the spotlight that there are almost no photos of him online, and Asai told him how Jobs actually ran Apple after the July 1997 return: ninety days from bankruptcy and down in every detail.¹ Chesky chose to install that posture rather than admire it.

The second anchor bolted the correction to the floor

Lutke ran the same spring in the same year, and the difference in what he did afterward is the reason this plate carries two anchors instead of one. In 2020 he spent stretches of sixteen-hour days reviewing every active project at Shopify and found a team in Toronto building supermarket features on the reasoning that "the supermarket industry is large and we should capture 1%."³ Nobody had told him the team existed. By the end of the reviews he had cancelled roughly sixty percent of everything in flight, and within a year he had turned over one hundred percent of the executive team.³ His verdict on the five years that grew the supermarket team is the sharpest self-diagnosis in the catalog. "I was cosplaying a public company CEO. That almost killed the company."³

Everything to that point is Chesky's move. The variant is what Lutke does with the correction once he has it: he does not hold the pace with his own body, he encodes it in structure so it cannot decay. He went to GitHub and started over. Shopify OS was Python and config files feeding a constraint solver that computed the org the company was supposed to have, which is how 5,500 inconsistently ranked job titles across 8,000 employees stopped being invisible.³ Major work now runs through a phase-transition checkpoint where a team trades accountability for autonomy in the open, so nothing accumulates unseen again.³ An internal podcast called Context records the reasoning behind decisions rather than the decisions, at twenty-three minutes an episode, twenty-three because someone computed the average Shopify commute and he refuses any number he cannot defend.³

He is equally exact about what must never be encoded. "Actively avoiding to write something down is sometimes the most important thing you can possibly do," because writing down what a company selects for publishes a cheat sheet, and optimizers will perform the signal without ever being the thing.³ Context is documented because context compounds. Criteria are guarded because writing them down is what rots them. A founder who gets that call wrong in either direction has engineered a fresh failure.

The archive kept these two on separate plates on the theory that the difference was tempo, Chesky continuous and Lutke once. The record falsifies that in both directions. Chesky's rooms leave permanent structure behind, teams that hold the standard after he walks out and never get the pace handed back. Lutke corrects continuously, and the machinery he built in 2020 exists precisely so the correction runs when he is not in the room. The real difference is smaller and more useful than tempo: it is where the correction is stored. One man stores it in his own presence and has to keep spending it. The other stores it in the building. The origin modes differ too, and the archive records the difference rather than smoothing it: Chesky chose this posture and can say what he was choosing to stop being, while Lutke's own file reads lost-and-recovered, because he had the answer early, performed his way out of it for five years, and was handed it back by a catastrophe he did not schedule.

The method transfers. The hunger does not.

The honest answer is that the method copies and the person underneath does not. Anyone can run the loop: pick one small surface that actually matters, get in the room, review the work often enough and long enough that a higher standard becomes reflex, then repeat next door. The tactics are explicit and portable, whiteboard material, the kind of thing that could be handed to someone in a completely different industry.

What does not port is the hunger. Chesky reviewed everything at 100 hours a week for years and tells the story now without a trace of regret. That is a temperament, not a technique, and specifically a designer's temperament. He says he feels like a designer more than a CEO,¹ and the stories he reaches for are coach stories, none more telling than John Wooden opening a player's first practice at UCLA with a full hour on how to put on socks.¹ Chesky heard the sock hour and got obsessed. That reaction is the filter: take away the obsession and "renovate one room at a time" degrades into a manager doing drive-by reviews, which is worse than absence. The spirit transfers to a founder who already has the hunger, and the hunger is not a framework that can be adopted on a Monday. That is what learnable in spirit means.

One man removed himself. The other removed the people.

Line up enough scaled founders and the same fork keeps appearing: the company gets big, the founder lands back in the product reviews, and the reviews either raise the work or bend it toward the boss. The taxonomy used to mark that fork by setting Chesky beside Daniel Ek and calling each of them right for his own archetype. That is a courtesy, not a distinction. A pair of plates that cannot be told apart on a live case is one plate with two names, and this pair sat untestable for as long as the concession stood.

So here is the test, put to the anchor of each plate within a few years of the other. When the founder is told, to his face, that his presence is degrading the team's output, who gets removed?

Ek's product head told him the team was spending the product reviews appeasing him rather than building, and that he was adding less value than he believed.⁴ Ek's first instinct was rage. Then he gave the claim three months, decided the verdict was correct, and never ran product again.⁴ He removed himself, permanently, and that removal is the coach.

Chesky heard the identical verdict at Airbnb and drew the opposite conclusion. "The initial attempt at founder mode created giant revolts. Everyone hated it because everyone thought I was micromanaging them."¹ He did not step back and he did not soften the pace to fit the objection. "Now, the people that hated it left."¹ He removed the people, and the standard stayed in the room.

The test is cheap to run, it fires on a specific dated event rather than on temperament, and it can come back wrong, which is what makes it worth writing down. A founder who hears that verdict and vacates the seat is a coach. A founder who hears it and replaces the room is a re-founder. It is the one moment where the two doctrines issue contradictory instructions and somebody has to be obeyed.

The re-founder's specific danger sits on the same fork, and it is easy to name and hard to see from the inside. Presence curdles into micromanagement, and the revolt is not proof of either reading. "One room at a time" also collides with the arithmetic of a single body: Chesky personally held the bar for years before the pace stuck, and nobody has measured how many rooms one founder can hold at that pressure before the method gives out. Lutke's answer to that arithmetic, storing the correction in structure instead of in himself, is the only version of this plate that scales past one person's hours. The whole discipline comes down to telling the difference between clearing an obstacle and being one, and nothing in the doctrine makes that call.

He took the dream literally

The dream compresses the archetype. Chesky leaves for ten years, comes home to a company he cannot recognize, and learns that the man who let it happen is the man who left. The re-founder is the founder who takes that dream literally and declines to run the experiment. Chesky does not wait ten years to come back. He never quite leaves. Whether that is wisdom or a cage depends on whether the founder would enjoy the work, and Chesky, by his own account, does. There are a lot of rooms in a hundred-billion-dollar house.