Airbnb says AI is helping it ship features faster as it tests a new search function

Airbnb has quietly become an AI-first company on the back end, even as it keeps front-facing chatbots at arm's length. On the company's second-quarter earnings call, CEO Brian Chesky laid out numbers that read more like a software factory than a hospitality app: features that used to take months to go from idea to launch now move 60% faster, and Airbnb shipped nearly 80% more features and improvements over the last six months compared with the same stretch a year earlier. Chesky credits AI tools for writing 60% of the company's code, a stat first floated earlier this year that now looks less like a headline and more like an operating principle.

Where this shows up matters more than the raw percentages. Search, sign-up, checkout, payments, and even host onboarding have all been touched by AI-assisted engineering, according to Chesky. That's the unglamorous plumbing of a travel marketplace, and speeding it up doesn't generate flashy demos, but it does compound. Faster iteration cycles mean Airbnb can test more ideas against real users without burning a year on each one.

On the consumer side, though, Chesky has been stubbornly cautious. He's said more than once that a generic chatbot bolted onto a travel app doesn't actually solve anything, and Airbnb's public-facing AI so far has stuck to safer ground: review summaries, listing highlights, that sort of thing. Now the company is finally testing something bigger — an AI search mode that users can toggle on, letting them type natural-language queries and get back a more visual, conversational results page. Chesky described it as titles that read like a chatbot wrote them, paired with real-time, personalized highlights once you land on a listing. Crucially, it's opt-in, not a forced replacement for the filter-and-scroll search people already know.

Customer support tells a cleaner AI success story. Airbnb's AI support bot launched in North America in 2025, expanded to more than 50 languages, and is headed toward voice calls later this year. Chesky says 45% of support cases that start with the bot now get resolved without a human ever stepping in, and that's translated into a 16% year-over-year drop in support cost per booking. Meanwhile the business itself is healthy — revenue climbed 17% to $3.6 billion for the quarter, with adjusted EBITDA up 21% to $1.3 billion — so none of this AI push is happening out of desperation.