Chesky Argues AI Agents Need an Operating System Built Just for Them

This week, Airbnb released an AI-powered search feature as part of its fall update, a notable shift for a company known for proceeding slowly and cautiously with AI. Brian Chesky, co-founder and CEO, has previously stated that a chatbot-only interface would be inadequate for travel discovery. With more AI now arriving on the Airbnb app, the company is leaving behind that earlier stance — though it is not heading toward a chatbot instead.

Over Zoom, Chesky made the case that Airbnb’s present search interface is merely a temporary stage. He said it will keep changing, moving toward something sitting between a chatbot and the first version the company shipped.

“I think that I’ve believed for a long time that a chatbot isn’t the right interface for e-commerce,” Chesky said. “I think what you’re seeing today is also not the endgame for e-commerce or for travel or shopping. I think that there’s going to be some interface that is, you know, I don’t want to call it a midpoint, but something between a chatbot and what you see in the first version we shipped.”

What the chatbot problem is

Two complaints underpin Chesky’s argument against chatbots. One of them is that chatbots present only a handful of options at any given moment, compelling users to make several turns before arriving at an outcome. He said this arrangement vexes browsing and shopping, since customers prefer to view a complete collection of options all at once.

The second grievance concerns the individual nature of chatbots. They function as a single-person interface rather than a shared one, which is a problem given that people frequently use these products alongside family and friends.

For the coming three to six months, the company’s work centers on studying interfaces that would allow multiple users to engage with “multiplayer” AI at the same time. That kind of system would let, for example, a group of people planning a trip together use it as a shared tool.

Why people want to look

Chesky referenced studies indicating that individuals get greater enjoyment from planning and looking forward to trips than they do from the journeys themselves. The takeaway is that the app should protect the delight of browsing, rather than taking it away.

“You know, a lot of people want some things … to be fast and easy, like ‘Book me a flight, I don’t want to look at it,’ ” he said. “With Airbnb, they enjoy looking at the product. Travel studies show that people actually derive more pleasure in planning and anticipating travel than they do traveling. So you don’t want to take the inspiration and the browsing and the library experience of dreaming away.”

The new shape of interfaces

Other companies are also questioning how AI interfaces should be built. New ways of presenting information are appearing within apps as AI tools keep changing. Young firms such as Monogram and, more recently, Wabi are testing “generative” interfaces — screens an AI puts together on the spot instead of ones a designer worked out ahead of time.

Chesky believes users will encounter both deterministic (pre-designed and predictable) and generative interfaces. The company itself takes a position that falls somewhere between these two approaches.

Preparing for agents

Agents are tools like Instinct and Muse that carry out tasks on a person’s behalf. Chesky said they will need robust software infrastructure on Airbnb’s end, and the company wants to make itself more agent-friendly. He sees them as potential lead generators.

“I don’t necessarily know if I believe in a world that there’s apps in the future, but I don’t believe in a world where the apps become data layers of a universal simple UI,” he said. “When you open your iPhone, there is a reason Airbnb’s app is a different design than DoorDash, Shopify, and Amazon. Our interfaces should be adaptive to the utility and the function that we’re doing. We want to make our app more agentic, and we want to build a lot more user interface controls.”

He also touched on the practical limits of agents. A person who uses one still wants to browse listings, message hosts, compare prices, verify their identity, use a map, and add other items. The agent either hands off to the app, or the agent application needs to build a more robust software developer kit to bring those app features to the person.

Interoperability and MCP

The plan includes letting agents talk to agents across the platform itself, along with agents from other apps. Chesky’s point is that apps haven’t always worked together unless companies agreed on a deal first.

On the other hand, agents are like “apps with faces,” and when they talk to each other, all apps can become interoperable. “I think of the Airbnb app as one day becoming an agent, you know, where basically the explore tab has an agent, our customer service has an agent, we’re developing another agent in another service area. Probably eventually there’s going to be a macro Airbnb agent. And that will probably be interoperable with MCP [a standard that lets AI agents connect with other software] to other agents,” Chesky noted.

The operating system gap

Chesky’s biggest argument is structural. He thinks the industry lacks a true operating system built for AI. Right now people are building AI apps for iOS, macOS, and Windows that are not “AI operating systems.” He thinks that in an ideal world, AI would be operable at the kernel level — the core of an operating system — and all components like agents would interact with each other.

“The problem with that model [of trying to become primary agent] is no one is developing a full operating system software developer kit,” he said. “I’ll give you an example. ChatGPT, at the beginning of last year, tried to do their little app store. And I told Sam [Altman]: If you want to be an app store like the iPhone, you need to have a software developer kit and an operating system like the App Store. Otherwise, none of our functionality is going to exist. And of course, it didn’t work very well.”

He also shared his own experience with Instinct and Muse. “I’ve used Airbnb on Muse and Instinct,” he said. “And it doesn’t work very well. I mean, even hotels don’t work that well, but Airbnbs work even worse.”

Where this goes next

Chesky’s view is that the current version of Airbnb’s search is a stop along the way, not the destination. The interface will evolve over time, moving toward something between a chatbot and the first version the company shipped.

The race, as Chesky sees it, is to be the “quarterback,” or primary AI agent that users go through for everything. But he argues that model won’t work until someone builds a full operating system software developer kit.

The argument here is that the industry requires an actual operating system for artificial intelligence. Until such a system exists, firms like Airbnb will continue developing interfaces that bridge the gap between the old and the new — a place where searching remains important, and where agents must pass control back to the app itself.

The question is whether anyone will build that operating system first.

Source material: “Brian Chesky interview: AI agents need their own operating system,” TechCrunch.

AI agents · Airbnb · Brian Chesky · chatbots · OS · travel tech