--- title: "How Google, Expedia, and Booking.com Will Respond to AI" description: Each incumbent is approaching AI differently and each is constrained by existing revenue models. Here is our analysis of their strategies and limits. canonical: https://nowah.xyz/blog/how-incumbents-respond-to-ai-travel lastModified: "2026-08-07T03:48:22.328Z" --- # How Google, Expedia, and Booking.com Will Respond to AI Each incumbent is approaching AI differently and each is constrained by existing revenue models. Here is our analysis of their strategies and limits. Every large travel company is investing in AI. The press releases are indistinguishable. "We are leveraging artificial intelligence to transform the travel experience." The actual strategies behind those press releases are wildly different, and the constraints each company faces determine what they can actually build. I follow these strategies closely because understanding what incumbents can and cannot do informs what we build and how fast we need to build it. Here is my read on where each major player is headed and where they will hit walls. ## Google: massive data, no booking ![Illustration for this section](https://pics.nowah.xyz/website-media/engineering-059-img-1.webp) Google has the best position in travel AI from a data perspective and the worst position from a business model perspective. [Google Flights](/blog/best-flight-booking-2026-ai-vs-google) already has extraordinary data. Flight [pricing across](/blog/currency-pricing-across-markets) every carrier. Historical price trends. [Connection quality](/blog/airport-intelligence-connection-quality) data. Airport delay statistics. Combine this with Google's AI capabilities and you get an AI travel assistant that could be genuinely brilliant at finding the right flights. The problem is that Google Flights does not book flights. It redirects you to the airline's website or to an OTA. Google is a search and advertising company. Its travel revenue comes from other companies paying to appear in Google's search results. Building a direct booking capability would cannibalize that revenue stream. Google's AI travel features reflect this constraint. They are search enhancements, not booking capabilities. AI-generated summaries of travel options. Smart suggestions based on search history. Predicted price trends. All useful. None of them close the loop to a booking. Could Google build a booking engine? Of course. They have the engineering talent and the capital. But the organizational decision to cannibalize search advertising revenue in travel is a strategic choice that requires board-level conviction. As of now, they have chosen to enhance search rather than replace it. The opportunity this creates for AI-native platforms is significant. Google will continue to be the starting point for many travel searches. But every search that starts on Google and ends on a separate platform to complete the booking is a friction point that a conversational AI agent eliminates entirely. ## Expedia: the chatbot-on-a-funnel approach Expedia was one of the first major OTAs to ship an AI chat feature, and their approach is instructive. The chatbot can help you think about your trip. It answers questions about destinations. It suggests dates based on your criteria. It can narrow down your options through conversation. But when you are ready to actually search for specific flights or hotels, you transition into Expedia's traditional search-and-results flow. This is the chatbot-on-a-funnel pattern. The AI handles the inspirational, pre-decision phase. The legacy funnel handles the transactional, decision-and-booking phase. The two systems sit side by side rather than being integrated. The architectural reason for this split is that Expedia's booking funnel was built over decades to handle billions of dollars in transactions. It is battle-tested, compliant, and reliable. Replacing it with AI-driven booking would require rebuilding the entire transaction pipeline with new reliability guarantees, new [error handling](/blog/error-handling-conversational-systems), and new compliance controls. No rational engineering leader would take that risk without a compelling business reason. The business reason might be coming, though. OTA [conversion rates](/blog/booking-conversion-rates-ai-agents) have been stuck between 2% and 5% for years. If AI-native competitors can demonstrate significantly higher conversion rates through [conversational booking](/blog/ai-travel-booking-conversation-first), that would be the business case Expedia needs to justify the rebuild. But right now, the chatbot-on-a-funnel approach is the safe bet, and large companies optimize for safety. ## Booking.com: hotel-first with AI polish ![Supporting diagram](https://pics.nowah.xyz/website-media/engineering-059-img-2.webp) Booking.com has taken a more thoughtful approach than most. Their AI trip planner does a reasonable job of helping users discover destinations and plan itineraries. The AI integration within hotel listings is genuinely useful, surfacing relevant reviews and highlighting amenities that match the user's stated preferences. But Booking.com has a structural limitation: they are primarily a hotel platform. Their flight booking capabilities are limited compared to dedicated flight OTAs. Their AI features reflect this bias. Hotel-related AI is good. Flight-related AI is basic. Multi-component trip planning (combining flights, hotels, and activities) is where the experience falls apart. For a user who wants to book a hotel in a city they have already decided on, Booking.com's AI is helpful. For a user who wants an AI to plan their entire trip from scratch, it falls short because the underlying platform was not built for end-to-end trip assembly. ## Hopper and Kayak: more nimble, still not native Hopper is the most interesting intermediate player. Their price prediction feature, which tells you whether to buy now or wait for a better price, is a genuine AI application (though it is more machine learning than generative AI). They have a younger, more mobile-native user base. Their app experience is cleaner than the legacy OTAs. But Hopper is still built around the browse-and-buy paradigm. The AI improves the existing flow. It does not replace it. You still search with structured inputs, browse results, and tap through a booking form. Kayak occupies a similar position. Good technology team. Interesting AI experiments. But fundamentally a meta-search engine that compares prices across OTAs. Adding AI to a meta-search product means better search, not different search. ## Why all incumbents are constrained The pattern across every incumbent is the same: existing revenue models constrain AI strategy. Google cannot build booking because it would cannibalize search ads. Expedia cannot replace the booking funnel because it would risk the transaction pipeline. Booking.com cannot expand beyond hotels because their supplier relationships and compliance infrastructure are hotel-specific. Hopper cannot abandon browse-and-buy because their price prediction feature depends on users making the final decision. These constraints are rational. Each company is optimizing for its current business. But in aggregate, they create a gap that AI-native platforms can fill. The user who wants to say "book me a trip to Japan" and have an AI handle everything from flights to hotels to activities to payments in a single conversation cannot get that experience from any incumbent today. Incumbents are slow to cannibalize themselves. That slowness is the window of opportunity for AI-native challengers. It will not stay open forever. But it is wide open right now, and the companies that build the best AI travel agents while the window is open will have a head start that is very hard to close. --- Nowah is an AI travel agent that searches and books real flights and hotels through conversation — no filters, no thirty open tabs. [Plan your next trip](https://app.nowah.xyz).