Airbnb is to push AI deeper into its core app and operations, CEO Brian Chesky told analysts, claiming that the company is already resolving about a third of customer support issues through an AI Assistant in North America, and is running live tests of AI-driven search for a small slice of traffic.
In Airbnb’s Q4 2025 shareholder letter, the company said its AI-powered customer support is now available for English, French and Spanish-speaking users across the US, Canada and Mexico.
On the earnings call, Chesky described the same push in operational terms, saying nearly 30% of English-language tickets in North America are handled by an AI agent today.
Support is the first measurable AI wedge
Chesky said Airbnb expects its AI agent to handle “significantly more than 30%” of customer support tickets within a year. He also said the company plans to expand the agent to “all the languages where we have live agents” and add voice so customers can “call and talk to an AI agent.”
Airbnb described the upside as both cost and service performance, citing faster responses and higher productivity for human agents using AI.
For a marketplace where disputes, cancellations and exceptions are routine, Chesky emphasized that high-quality automated support is a driver for repeat use in the marketplace.
AI search is live, but Airbnb says the interface is the hard part
On the call, Chesky said AI search is “live to a very small percent of traffic right now” and that Airbnb is iterating quickly rather than planning a single major launch.
He also argued that commerce search is difficult for chatbots because travel is highly visual and comparison-driven, which pushes Airbnb toward redesigning the user interface instead of simply dropping a text chatbot on top of listings.
In the shareholder letter, Airbnb described early AI search tests as focused on letting guests describe what they want more naturally and ask questions about the listing and location, with the aim of evolving into a broader search and planning experience that “extends through the trip.”
Ads are being positioned as “later,” after the search form factor stabilizes
One strategic thread Airbnb is making explicit is sequencing. Chesky said the company wants to design AI search first and only then, if it introduces sponsored listings, design that ad unit for the AI search form factor.
That framing suggests Airbnb is treating AI search as a product transition with monetization implications, not as a feature bolt-on.
A new CTO and a “data plus workflows” moat argument
Airbnb’s AI rebuild is tied to leadership changes. Airbnb’s newsroom announcement said Ahmad Al-Dahle joined as CTO after roles at Apple and Meta, where Airbnb says he founded Meta’s Generative AI group and led the launch of Llama.
Chesky also referenced his background on the call while describing the company’s intent to become more “AI-native.”
Chesky’s competitive claim is that generic AI assistants can replicate model access, but not Airbnb’s proprietary trust and marketplace layers.
He cited 200 million verified identities and 500 million proprietary reviews and said chatbots cannot message hosts, which he said 90% of guests do.
He also pointed to Airbnb’s broader infrastructure, including payments, customer support and insurance as capabilities that are harder to recreate quickly.
The distribution bet: AI chatbots as top-of-funnel, not a replacement
Chesky told analysts that chatbot platforms will behave like search and that Airbnb has seen traffic from chatbots convert at a higher rate than traffic from Google.
At the same time, he argued that the underlying models are broadly available to competitors, so differentiation shifts to specialization, post-training and connecting AI to Airbnb’s own interactions and workflows.