Airbnb to Test AI Search as Code Is 60% AI-Generated

Airbnb will begin testing an AI-powered search feature later this year, offering users a natural language alternative to its traditional filters and maps. The development comes as the company reports that AI now generates 60% of its new code, helping to accelerate feature development by nearly 80% compared to the same period last year.

The company’s cautious approach to consumer-facing AI—avoiding chatbots in favor of visual, personalized discovery—reflects a broader strategy to integrate the technology deeply into its internal operations before reshaping the user experience. During its second-quarter earnings call, co-founder and CEO Brian Chesky outlined how these dual-track investments are now producing tangible results.

What Happened

Airbnb is preparing to test a new AI search function that operates alongside its existing interface. Unlike the current system of filters and map-based browsing, the new feature allows users to type natural language queries and receive visual results with AI-generated titles and personalized highlights on listing pages. The key is a toggle, letting users choose between the familiar experience and the new AI-driven alternative.

This comes as the company reported strong Q2 2026 financials, with revenue reaching $3.6 billion—a 17% year-over-year increase—and adjusted EBITDA of $1.3 billion, up 21%.

Behind the scenes, AI is playing a much larger role. Chesky confirmed that the company has reduced the time from concept to launch by up to 60% across key initiatives, and the total number of features shipped in the first half of 2026 is up nearly 80% year-over-year. These improvements span search, sign-up, checkout, and payments, as well as host-facing tools like faster onboarding flows.

Airbnb has already deployed an AI-powered customer support bot across North America, expanding it to more than 50 languages this year. The company reports that nearly 45% of customer issues that begin with the AI agent are resolved without human intervention, contributing to a 16% year-over-year drop in support costs per booking.

Why It Matters

For a platform of Airbnb’s scale—with millions of listings, guests, and hosts—the speed at which it can iterate is a competitive moat. The 80% increase in features shipped suggests that AI is not merely an efficiency tool but a fundamental change to how the company builds product. Where traditional software development might take months to test and refine a new feature, Airbnb is now operating on a timeline that allows for rapid experimentation and iteration.

This is particularly significant for the travel industry, where user expectations shift with economic conditions and seasonal patterns. Faster development cycles mean Airbnb can respond to trends—such as the rise of “workations” or the demand for pet-friendly stays—more quickly than competitors who rely on older development methodologies. The reduction in support costs also signals that AI can handle the volume of routine inquiries, freeing human agents to tackle more complex issues, which could improve overall customer satisfaction.

However, the consumer-facing rollout is deliberately measured. Chesky has previously expressed skepticism about chatbots as the primary interface for travel planning, arguing that the discovery process is inherently visual and intuitive rather than conversational. The toggle approach allows Airbnb to test AI search with a subset of users without disrupting the experience for those who prefer the existing system. This “parallel testing” strategy suggests Airbnb is hedging its bets, collecting real-world data before committing to a broader rollout.

Background and Context

Airbnb’s approach to AI has been notably different from that of many tech companies. While competitors like Expedia and Booking Holdings have experimented with generative AI for trip planning, Airbnb has been vocal about its reluctance to simply add a chat interface. The company has instead focused on using AI to improve the underlying search and discovery engine, as well as to automate backend processes.

This isn’t the first time Airbnb has introduced AI features. In recent years, the company launched review summaries and listing highlights, both of which rely on AI to distill large amounts of user-generated text into digestible insights. The new AI search test builds on this foundation, extending AI’s reach into the core discovery experience.

The company’s internal adoption of AI has been more aggressive. The claim that AI writes 60% of its code—first made earlier this year—positions Airbnb alongside other tech giants like Google and Microsoft in leveraging AI for software development. This shift has implications beyond speed; it also changes the nature of engineering work at the company, requiring developers to focus more on review, integration, and system design rather than writing boilerplate code.

Key Details

The AI Search Toggle: The new search feature will be accessible via a toggle, allowing users to switch between traditional and AI-driven modes. In AI mode, users can type natural language queries—for example, “a beachfront villa with a pool for a large family”—and receive visually rich results. The titles of these results will be AI-generated, and the listing highlights will be personalized in real time. This approach mimics the conversational style of a chatbot without adopting the text-heavy interface, aligning with Chesky’s stated preference for visual discovery.

Customer Support Automation: Airbnb’s AI support bot now handles 45% of initial customer issues to completion. The bot was launched in North America in 2025 and expanded to over 50 languages in 2026, with plans to support voice calls later this year. The 16% reduction in support costs per booking indicates that the bot is not only handling volume but doing so cost-effectively.

Development Velocity: The 60% reduction in time from concept to launch and the 80% increase in features shipped are the most striking figures from the earnings call. These improvements are attributed to AI-assisted coding, which allows developers to generate code faster, test more quickly, and iterate on features with shorter feedback loops. The company has not disclosed which aspects of the development lifecycle have been most affected, but the broad impact across search, sign-up, checkout, and payments suggests that AI is being integrated into multiple parts of the stack.

Original Editorial Analysis: The Soft Launch Strategy

Airbnb’s toggle-based AI search test reflects a deliberate and perhaps unusual strategy: it’s treating AI as an optional enhancement rather than a mandatory upgrade. This stands in contrast to the approach taken by many tech companies, where new features are often rolled out as defaults to encourage adoption and gather data.

There are two plausible explanations for this cautiousness. The first is practical: travel planning is a high-stakes, high-information activity. Users have established mental models for how to find a place to stay, and disrupting that with an unfamiliar interface risks frustrating those who are not interested in change. The toggle allows Airbnb to gather feedback from a self-selecting group of early adopters without alienating the broader user base.

The second explanation is more strategic: Airbnb may be using the toggle to set clear expectations about AI’s role. By presenting AI search as an option rather than a replacement, the company implicitly acknowledges that the technology is not yet ready to fully replace traditional search. This transparency—whether intentional or not—could build trust with users who are skeptical of AI’s ability to understand complex, subjective queries like “a cozy cabin with a fireplace and mountain views.”

This approach also gives Airbnb a built-in control group for measuring the effectiveness of AI search. By comparing conversion rates, booking times, and user satisfaction between the traditional and AI-driven interfaces, the company can quantify the value of its AI investments. If the AI toggle proves successful, Airbnb can then make it the default with data to back up the decision.

In short, the toggle is not just a usability feature; it’s an experiment and a confidence-builder. It reflects a company that is aggressively using AI to build products but remains cautious about how that technology is presented to consumers.

Industry Implications

Airbnb’s development velocity gains have competitive implications. Traditional online travel agencies (OTAs) like Booking.com and Expedia have also been investing in AI, but Airbnb’s ability to ship features 80% faster could allow it to outpace them in areas like user experience, pricing tools, and host management.

For hosts, the faster development cycle means more frequent updates to the platform’s tools, potentially including AI-powered pricing recommendations, dynamic listing optimization, and automated guest communication. The expanded language support for the customer support bot also benefits international hosts who may rely on Airbnb’s automated tools to manage their bookings.

For guests, the AI search test represents the beginning of a more personalized discovery experience. The personalized highlights generated in real time could make it easier to find properties that match specific preferences, reducing the time spent scrolling through irrelevant listings. However, the success of this feature will depend on the quality of the AI’s recommendations and its ability to handle nuanced queries.

Related Developments

Airbnb is not alone in exploring AI for travel search. Expedia has tested a ChatGPT-powered trip planner, and Google’s Travel offerings have incorporated AI for flight and hotel recommendations. However, Airbnb’s focus on visual results and personalized highlights distinguishes its approach from the text-heavy interfaces of its competitors.

The company’s internal adoption of AI for coding also echoes broader industry trends. Microsoft, Google, and Amazon have all reported using AI to accelerate development, and startups like GitHub (with Copilot) and Replit have built entire businesses around AI-assisted coding. Airbnb’s reported 60% figure places it among the more aggressive adopters of this technology.

Airbnb’s AI journey is a story of two speeds: rapid internal adoption and measured consumer rollout. With 60% of its code now AI-generated and an 80% increase in feature shipping, the company has demonstrated that AI can fundamentally reshape product development. At the same time, the toggle-based AI search test shows a careful, iterative approach to consumer-facing features, reflecting a respect for the complexity of travel planning and the habits of its users.

The most significant takeaway is that Airbnb is using AI to solve two distinct problems: making its engineers more productive and making its product more personalized. The success of the first effort is already reflected in the financial results and the pace of feature releases. The success of the second will depend on how well the AI search toggle performs in testing and whether users find genuine value in a more conversational, visual discovery experience. For now, Airbnb is betting that a soft launch is the smartest way to learn what its users actually want.