Airbnb uses AI to accelerate product delivery and test search
Airbnb reports faster product releases with AI
Airbnb says artificial intelligence has reduced the time from concept to launch by as much as 60% across some key initiatives. During the company’s second-quarter earnings call, co-founder and chief executive Brian Chesky said the company also shipped nearly 80% more features and improvements than in the same six months a year earlier.
The company had said earlier this year that AI writes 60% of its code. Chesky said the technology is helping teams build, test and iterate more quickly, with work spanning search, sign-up, checkout and payments. Airbnb has also introduced host-facing changes, including a faster onboarding flow.
Airbnb reported revenue of $3.6 billion for the quarter ended in June, a 17% year-over-year increase. Adjusted EBITDA rose 21% to $1.3 billion.
Optional AI search moves into testing
On the consumer side, Airbnb has taken a more measured approach. Its existing AI features have largely included review summaries and listing highlights, while Chesky has argued that simply placing a chatbot interface in front of travel planning would not suit the use case.
The company will now begin testing AI search focused on search, discovery and support. Rather than replace the existing search and filter experience, Airbnb plans to add a toggle that allows users to switch to AI search and enter requests in natural language.
Chesky said results would be presented visually. Titles in responses could be AI-generated and conversational, while listing-page highlights could be generated in real time and personalised to the user. This approach keeps the familiar search path available while testing a different way to explore inventory.
Support automation delivers a measurable cost result
Customer support is a major back-end deployment area. Airbnb launched its AI-powered bot in North America in 2025 and has expanded it to more than 50 languages this year. The company plans to make the service available for voice calls later this year.
Nearly 45% of customer issues that begin with Airbnb’s AI agent are completed without human intervention, the company said. Support cost per booking is down 16% year over year. The operational focus reflects a broader industry discussion around AI agents and emerging sales models and the business models emerging around AI agents.
For businesses, Airbnb’s figures underline the value of judging AI deployments through concrete operating measures: development cycle time, release volume, issue resolution and cost per transaction. Optional interfaces can also provide a controlled way to test new AI-led customer journeys without forcing an immediate replacement of established workflows.