Chesky said 50 percent of Airbnb’s US users are already using the AI bot for customer service, adding that the company plans to make it available to all US users this month.
On the analyst call, he noted: “One thing I’ll say about AI is that it really does make the customer experience easier… it has already reduced the number of people who need to contact a human agent by 15 percent.”
Last year, the company told TechCrunch it was testing the technology in limited fashion and only for certain queries. “I think the potential for AI in the business world is enormous. We’ve been thinking about how AI will change the consumer-level experience over time,” Airbnb co-founder Nathan Blecharczyk told TechCrunch at the time.
Unlike OpenAI, Google, Perplexity, and the many startups building AI agents (tools that execute tasks on behalf of users), Airbnb appears to be taking a more cautious approach. Chesky said in February that the company would apply AI to customer service first, then travel planning or flight booking later, because he believes the technology is still early.
Meanwhile, competitors Expedia and Booking are investing aggressively — launching AI trip planning, travel itineraries, and real-time travel updates.
Airbnb reported Q1 total revenue of $2.27 billion, up 6 percent year over year. However, the company’s revenue forecast for the quarter came in slightly below analyst expectations, citing a slowdown in travel demand as global tariff wars dent sentiment and curb discretionary spending.
The 50-percent adoption figure also provides a rare data point on AI customer-service deployment in practice. Most coverage of AI support focuses on vendor claims or pilot programs; Airbnb’s disclosure covers a live production system serving half of a national user base at a company processing millions of stays per quarter. At that scale, the 15 percent human-contact reduction is not a projection — it is a measured operational delta, which makes it one of the most credible data points in the AI customer-service market. The implicit math: if Airbnb’s US user base generates millions of support contacts per quarter, a 15 percent reduction represents significant avoided labor cost, and the bot’s coverage expanding to 100 percent of US users will roughly double that saving.
The staged approach also reflects a lesson the industry has been slow to learn: the best AI customer-service bots are not the most ambitious ones. They are the ones scoped tightly enough that the model’s failure modes — hallucinated policies, invented refund rules, misread emotional context — cannot cause real harm. Airbnb’s support use case has naturally bounded inputs (booking references, structured issue categories), well-defined resolution paths (refund, credit, rebooking, escalation), and a human fallback that users expect and accept. The travel-agent use case that Expedia and Booking are racing toward has none of those guardrails: an AI that books the wrong flight or plans an impossible itinerary creates costs far exceeding the support savings. Chesky’s sequencing instinct — support before booking — is the same instinct that made Airbnb successful as a company: prove the model works at small scale before extending it to high-stakes transactions.
The revenue miss in the same earnings call adds context to why AI efficiency matters now. A 6 percent growth rate with a below-expectations forecast means Airbnb is managing costs in a slowdown, and the AI bot’s 15 percent contact reduction is exactly the kind of margin lever that matters when growth is expensive. The timing is not coincidental: companies deploy AI cost savings most aggressively when revenue growth disappoints, and the chatbot’s expansion to all US users — announced alongside a soft forecast — suggests operational efficiency is doing work that user growth currently cannot.
Why Airbnb’s Caution May Be the Smart Play
The 15 percent reduction in human-agent contacts is the metric that should stop every customer-experience leader mid-scroll. It means the AI bot is not just deflecting easy queries — it is resolving enough real issues to keep one in seven people from needing a human at all, at a scale of 50 percent of the US user base. In an industry where customer support is a major cost center (Airbnb handles everything from lockouts to refunds to neighbor disputes), a 15 percent contact reduction translates directly to operational savings that compound with every new user the bot covers.
The contrast with Expedia and Booking is the strategic story. The competitors are racing to build AI travel agents — tools that plan itineraries, book flights, and manage trips end to end. Airbnb is deliberately NOT doing that yet, confining AI to the customer-service use case where the input is a complaint (bounded, text-based, with structured data about the booking) and the output is a resolution (refund, rebooking, information). That is the lowest-risk, highest-certainty AI deployment in travel: the task space is narrow, the success criteria are measurable, and the downside of a mistake is a misrouted email rather than a botched flight booking. Chesky’s sequencing — support first, booking agents later — is a product strategy that treats AI maturity as a prerequisite rather than an assumption. Whether that caution costs Airbnb competitive ground against Expedia’s and Booking’s more aggressive rollouts is the open question; the 15 percent contact reduction suggests the conservative approach is at least generating measurable results where it counts — and measurable is the operative word in a market full of AI claims that resist measurement.
