Airbnb’s CEO is preparing to redesign the company around artificial intelligence, even though he starts from an uncomfortable idea: software ages quickly, and an interface that looks perfect today can feel outdated ten years from now.
The two ideas seem to point in opposite directions, but Chesky sees them as part of the same problem. That contradiction helps explain how one of the few designers to end up running a Fortune 500 company thinks.
In December 2020, Airbnb went public and reached a valuation of US$100 billion. But Chesky remembers feeling worse the next day than he had before.
He woke up, put on sweatpants and joined a meeting on Zoom. It was the middle of the pandemic and, despite everything that had happened the day before, his routine was exactly the same. That was when he realized that if a US$100 billion valuation did not change how he felt, US$200 billion would not either.
Some time later, the then U.S. surgeon general asked him what he thought was the leading cause of death in the country. Chesky first answered cancer, then heart disease. The answer he got was different: loneliness.
That was when, he says, he realized he had spent years feeling quite alone and that work had become an addiction.
No steering wheel
Professionally, that crisis had started earlier, and the story is fairly well known.
Around 2019, with seven thousand employees, Chesky woke up one day and felt that he no longer recognized the company he was running. He compares it to driving a car without a steering wheel: he would say left and the company would go right. The problem was not that he had lost control. Nobody had it.
Then the pandemic arrived and Airbnb lost 80% of its business in eight weeks. That ended his hesitation. He reviewed virtually everything, worked one hundred hours a week for two or three years and never really let go again.
Out of that came what Paul Graham later called founder mode. Chesky explains it this way: at the beginning, you have to stay very close to the team and only gradually step back. If you delegate too early, by the time you want to correct something, it is often too late.
He compares it to learning golf. An instructor needs to watch you swing hundreds of times before leaving you on your own. If you learn the wrong way from the beginning, it is not enough to understand what you were doing wrong. You also have to undo what your body has already learned.
At first, he admits, practically everyone at the company hated it and thought he was micromanaging them. Those who could not accept that way of working eventually left.
Now he says he is entering another phase that he still does not quite know how to name. For the moment, he calls it “founder mode with AI.” He expects it to be even more intense because information, analysis and answers can now be available almost immediately.
Until now, founder mode meant roughly thirty-five hours of meetings a week, usually in groups, with the entire chain of command in the room and Chesky speaking last. It was how he found out what was actually happening inside the company. With AI, he believes much of that will no longer need to happen in a meeting.
The biggest change may be in middle management. Chesky believes managers who do nothing but manage people will become less valuable and that every leader will have to remain involved in the work itself. An engineering manager has to code, a legal leader has to read court rulings and a design director has to design. For Chesky, a manager cannot simply manage people; they have to stay close to the actual work. He often gives a rather unusual example: the Catholic Church has operated for two thousand years with only four layers of hierarchy.
Ten people
The other part of the method produces more tangible results.
Chesky assembled a team of ten or twelve people —designers, engineers, a couple of product people and data scientists— and gave them one thing to improve: conversion, meaning how many people who search eventually make a booking. He ran it like a startup inside Airbnb. First they fixed what was broken, then added improvements, and only after that did they rethink the entire user journey.
In the first year, that team generated about US$200 million in additional value. In the second, between US$400 million and US$500 million. They later repeated the formula with pricing and other specific problems across the business.
From that came a rule Chesky now uses when launching new businesses: one, ten, many. Start in one market, expand to ten and, once it works, scale it. He says it took him sixteen years to launch Airbnb’s second product because for a long time he tried to do the opposite: launch in one hundred cities from day one.
Behind that idea is something he learned at Y Combinator: early on, it is better to have one hundred people who truly want your product than the lukewarm attention of one million. His way of putting it is simple: before trying to heat an ocean, start with a bathtub.
If there is one thing Chesky does differently from many other CEOs, it is hiring.
He does not like the usual recruiting process. He argues that hiring a search firm, receiving fifty résumés, interviewing five people and then choosing among three finalists does not necessarily lead to the best candidate. He prefers meeting people continuously, whether there is an open job or not, and asking each person for the names of the three best professionals they have worked with. He also suggests working backward from the result: if you see an extraordinary advertisement, find out who made it.
At Airbnb, he still participates directly in hiring the company’s top two hundred people. He admits that this is unusual. His day often begins and ends with a call to the recruiting team. He spends two or three hours a day on it.
His logic is fairly simple: if you choose the right people from the beginning, you do not need to spend all your time managing them later. Chesky says the best people require very little supervision.
Now he is trying to answer another question: what will still have value when technology changes everything else?
His own profession does not come out particularly well in that comparison. Buildings, Chesky says, age well. Over time, they can even gain character.
Hardware ages reasonably well too. Software is different: an interface that looked excellent ten years ago can now feel completely outdated.
His conclusion is that the Airbnb app will not last forever and that, eventually, apps themselves may disappear and be replaced by agents. What does endure, he says, is the community, the brand, the principles and the identity. That is why, at one point, he told the company they were not building a product but a community.
He also recognizes the limit. Close to US$100 billion a year flows through Airbnb, and many people depend on that income. Experimenting there is not the same as doing it in a garage, because one bad decision can affect thousands of people. That is why Chesky tests the riskiest ideas in separate teams, where they can work without disturbing the core business. The question is what they would build if they had to create Airbnb again from scratch.
The escalator
There is a story Chesky often tells to explain the idea.
When he was a child, the shopping mall near his home had an escalator that seemed enormous to him, one of the biggest in the country. In his memory, it was almost like the Obelisk in Buenos Aires. Years later, he returned to the same place and was surprised to see that it was a fairly ordinary escalator, much smaller than he had imagined for years.
Over time, he realized that he had seen the top of the ladder in much the same way. From below, it looked a kilometer away. Now that he has reached it, he says, he has realized that he is only a few inches higher than everyone else.
