Airbnb now writes 60% of its code with AI and has cut the time from feature concept to launch by as much as 60%, CEO Brian Chesky said on the company's second-quarter earnings call. The shift has translated into an 80% year-over-year jump in features and improvements shipped over the first six months of 2026. Revenue for the quarter ended in June rose 17% to $3.6 billion, and adjusted EBITDA climbed 21% to $1.3 billion.
The productivity claim is the sharpest number any large consumer platform has put on internal AI adoption this year. Chesky said the gains show up in search, sign-up, checkout, and payments, along with host-side tooling like a faster onboarding flow.
What Airbnb has notably not done is ship a chatbot to travelers. Chesky has argued for more than a year that a conversational box is the wrong interface for booking a stay, and the company has confined consumer-facing AI to narrow surfaces — review summaries and listing highlights — while building the harder pieces underneath.
Key facts
- 01AI now writes 60% of Airbnb's code, and has cut concept-to-launch time by up to 60%.
- 02Airbnb shipped nearly 80% more features in the first half of 2026 than in the same period a year ago.
- 03The AI support agent handles 45% of customer issues without human intervention, cutting support cost per booking 16% year-over-year.
- 04Q2 revenue rose 17% year-over-year to $3.6 billion, with adjusted EBITDA up 21% to $1.3 billion.
- 05Airbnb will begin testing an opt-in AI search toggle with natural-language queries and AI-generated, personalized listing highlights.
That posture is about to change, carefully. Airbnb will begin testing an AI-powered search experience, but as an opt-in toggle rather than a default replacement for the existing search-and-filter flow. Users who switch it on can type natural-language queries and get a visual results page rather than a chat transcript.
The description Chesky gave suggests Airbnb is trying to keep the visual grammar of the current app while letting the model do the heavy lifting behind the scenes. Titles in the results are AI-generated, listing highlights are personalized in real time on the product page, and the interface still looks like Airbnb rather than a text window.
That is a deliberate design choice. Travel search punishes chat interfaces because users need to compare options side by side, and long conversational threads make that hard. A toggle lets Airbnb A/B test genuine demand without disrupting the millions of users who prefer the filter-based flow they already know.
Customer support is the area where Airbnb has gone furthest. The company launched its AI support agent in North America in 2025, expanded it to more than 50 languages this year, and plans to bring it to voice calls later in 2026. Nearly 45% of customer issues that begin with the AI agent now close without any human handoff.
The financial signal is clear: support cost per booking is down 16% year-over-year. That is the kind of margin lift that changes how a marketplace models its operating leverage, because support has historically scaled roughly with bookings. Breaking that ratio while bookings grow is the entire point of deploying AI in the back office.
The bigger question is whether the 60% code-generation figure translates into durable product velocity or just short-term throughput. Shipping 80% more features in six months is a headline number, but consumer platforms live and die on how well features land with users, not how quickly engineers push them out. Airbnb has not disclosed feature-level engagement or retention data tied to the AI-built releases, and the AI search rollout is still a test rather than a launch.
For the AI market, Airbnb is now the clearest public data point on what happens when a mature consumer platform routes most of its engineering through AI-assisted coding. The revenue and margin numbers say the internal bet is working; the deliberate pace of the consumer rollout says Chesky still does not believe a chatbot is the answer for travel. If the search toggle converts, expect other marketplaces — where filter-based discovery has been the standard for two decades — to follow the same opt-in pattern rather than the chat-first one that has dominated AI product launches so far.
Working on something we should cover, or seeing a story we missed? Send leads, documents, or feedback to hello@aichatdaily.com. For sensitive tips, see our secure tips page for Signal and PGP options.
Spotted an error? Email hello@aichatdaily.com with the URL and the issue, or read our full corrections policy.
