Picture a scene in the bustling streets of Tokyo at 3 AM, where your AI assistant is engaged in a lively conversation in the Osaka dialect with an izakaya owner, finalizing the last traditional room for your stay. Meanwhile, the latest volcanic activity updates from Bali prompt it to seamlessly adjust your stargazing itinerary for the following morning. At this moment, all you have to do is relax in your overwater villa in the Maldives, waiting as the charming hues of your snorkeling gear are automatically billed to the best exchange rate negotiated by AI. This is not a mere sequence from a sci-fi movie; rather, it is a glimpse into the evolving landscape of travel redefined by advancing artificial intelligence.

The emergence of DeepSeek has ripped open the seams of change in the travel industry. While many enthusiasts marvel at AI-generated cherry blossom viewing guides in Kyoto, seasoned travel veterans are sensing a more profound storm brewing. It appears that these AI agents, navigating across bytes, are attempting to slice through the services offered by Online Travel Agencies (OTAs) using algorithmic precision. Their goal is to reshape the entire industry into manageable data modules, paving the way for users to experience a more integrated, efficient one-stop travel service.

As a form of "intelligent agency" technology, AI agents simulate human behavior—executing complex tasks and delivering comprehensive results. Take Google's Duplex AI, for instance, which is capable of single-handedly making restaurant bookings and hotel inquiries with a near-human level of natural conversation. This innovation signifies a shift whereby AI moves from a passive responder to an active agent, fundamentally transforming the underlying logic of the travel service sector. Industry analysts suggest that as AI agents continue to develop, they might eventually connect users directly with service providers, circumventing the entrenched structure of OTAs.

The question that arises is whether AI agents will serve as a replacement for OTAs or as a complementary service. At the core of AI agents’ capabilities is their ability to rapidly consolidate and analyze vast amounts of travel data—spanning hotels, flights, attractions, and more—to offer personalized, seamless service from start to finish. The potential of bypassing the OTA platform's usual processes allows AI agents to integrate all necessary travel components within a single interface. This not only connects disparate aspects of travel—transportation, lodging, and local services—but also eliminates the frustrations of comparing prices and creating multiple user accounts across platforms. For example, Hopper, a travel technology firm, leverages AI to predict airfare fluctuations with a remarkable 90% accuracy, far outpacing similar features on OTA platforms like Expedia which often lag by a day or two. Such continuous service provided by AI agents is something traditional OTA platforms struggle to accomplish.

Moreover, AI agents learn from ongoing user behavior, constructing a dynamic “personal footprint database” that serves as a foundation for increasingly accurate travel recommendations. Consider the startup Layla (formerly known as Mighty Travels); its AI capabilities analyze users’ social media data, past trip history, and real-time preferences to automatically curate comprehensive travel itineraries that include flights, accommodations, attractions, and even unique local experiences—all while connecting directly to suppliers for bookings. This application of AI agents could very well signal a burgeoning trend toward high-end customized travel services.

Distinctively, unlike OTAs which primarily cater to consumers, AI agents can provide real-time data and market trends (such as predicted foot traffic and consumption preferences) back to businesses. Therefore, businesses can adapt their operational strategies based on these insights.

Take, for example, the hotel industry’s application of AI agents; these digital entities can dynamically adjust room rates based on live occupancy rates and surrounding activity data, bypassing OTA commission fees and enhancing hotel profit margins. Furthermore, through the utilization of AI agents, businesses, including hotels and tourist attractions, could evade the costly commission models traditionally offered by OTAs, a significant incentive driving businesses to favor AI solutions over OTAs.

Nevertheless, it’s important to recognize that, at this point, AI agents are not equipped to fully replace OTAs:

Firstly, OTAs boast robust networks of suppliers and a vast user base. For example, Booking.com owns over 28 million lodging options globally, 70% of which are exclusive partnerships built over two decades of grassroots development. For AI agents to efficiently connect smaller hotels, they must navigate challenges related to technical standardization and trust within commercial relationships. Likewise, Airbnb has explored automatic agreements with property owners but still relies on a human team due to legal concerns and cultural complexities.

Secondly, user trust in established OTA brands, derived from years of operation and accumulated reputation, cannot be overlooked. According to Statista’s surveys, 68% of users prefer OTAs over AI solutions when changes arise in travel plans, favoring the superior customer service and insurance mechanisms traditions OTAs have established. During the early days of the pandemic, for instance, Ctrip allocated over 1.2 billion yuan in ticket refunds—a feat not conceivable for AI agents.

Thirdly, OTAs currently offer comprehensive assurance services that outperform those available through AI agents. They have constructed all-encompassing protection measures from booking through to after-sales. Services like Tongcheng Travel's “Worry-free Itinerary” and “Worry-free Check-in” automatically coordinate refunds and price compensations in case flight delays lead to hotel no-shows. Presently, AI agents primarily focus on data integration, still relying on third-party services for dispute resolution and resource coordination.

Given these three factors, it becomes evident that, despite the immense potential of AI agents, the protective barriers of OTAs’ resources and the entrenched ecosystem pose significant challenges that are unlikely to be surmountable in the near term. The travel industry’s operational model will, for the time being, remain largely dependent on OTAs.

OTAs are wielding AI as a defensive weapon to maintain their throne.

While AI agents, as an emerging concept, do not yet pose a mortal threat to traditional OTAs, recent developments reveal that OTAs are not blind to the potential of AI agents. Companies like Ctrip, Meituan, Qunar, and Tongcheng have begun to establish long-term AI strategies. On February 17, Ctrip announced a series of high-level changes, appointing Chen Gang as Chief Product Officer tasked with formulating AI product strategies—a move widely interpreted as Ctrip's continued investment in technology and exploration of new tech applications within travel. Similarly, Meituan CEO Wang Xing openly stated the company's dedication to embracing AI and big data technologies, eyeing the potential of drone delivery and automation. Moreover, the AI model “Chengxin,” co-developed by Tongcheng Travel and Tencent, has integrated AI into its app for providing intelligent travel planning and accommodation solutions.

Looking ahead, the convergence of AI agents and OTAs seems inevitable: OTAs can harness AI technologies to boost efficiency, effectively embedding AI as the foundational infrastructure of their services. In parallel, independent AI companies could focus on niche scenarios (like business travel or outdoor adventures), fostering a collaborative ecosystem of “large platform plus specialized tools.”

Nevertheless, AI agents encounter prominent challenges surrounding data security and profitability.

To begin with, the current application of AI agents raises critical questions about data security and accountability dynamics. Notably, in 2022, the EU halted the Italian AI travel assistant TripWhistle for “violating GDPR,” as it failed to inform users how their data would be used for commercial engagements. As an application that combines user privacy with business data, deploying these AI agents locally may jeopardize their performance; conversely, if operated online, how can data security be ensured? This situation necessitates a careful balance on the part of AI service providers.

Furthermore, the path toward commercialization for AI agents remains uncertain. For instance, the “AI Tour Guizhou” mini-program was a collaborative venture between the government and OTA, indicating that future AI agents may rely upon OTA’s foundational data to receive operational subsidies from local governments, a technological upgrade of traditional methods.

Should AI agents pursue independent profitability, they might replicate the commission models that OTAs utilize. The startup JourneyMate endeavored to charge users a subscription fee but faced a dismal payment conversion rate of merely 3%, ultimately redirecting its focus toward B2B by providing AI customer service systems to hotels. This scenario exemplifies AI agents’ precarious position: engaging directly with consumers risks being swallowed by OTAs, effectively bolstering their monopolistic power. In contrast, B2B engagements reduce AI agents to technology suppliers, akin to selling “usage rights,” integrated within the products or services of other enterprises for licensing fees.

These two challenges are not insurmountable but reflect AI agents’ critical quest to avoid becoming “the next OTA.”

Similar to OTA services, AI agents also require a platform to aggregate and consolidate data. Should AI agents evolve into monopolistic intermediaries, they revert to the role of the new OTA. Google previously explored integrating maps, Gmail, and search data for travel planning and payments. If successful, Google could emerge as the first AI titan to disrupt OTAs, albeit exposing itself to monopolistic scrutiny—such as the antitrust investigations initiated against Google’s travel services in France in 2023. Conversely, while Amazon’s Alexa could have unified Booking.com, Uber, and other services, to avoid competition with partners, Alexa merely serves as a “gateway,” redirecting traffic back to OTAs, which can be seen as a form of compromise.

Ultimately, if AI agents evolve into entities that dominate user and business information but mirror the operational frameworks of traditional OTAs, it signifies a troubling scenario where the slayers become the dragon, a situation neither consumers nor businesses wish to witness. The capability of AI agents to achieve decentralization, avoiding the pitfalls of becoming monopolistic intermediaries, whilst effectively compiling prior scattered applications and platform data, will determine their future potential as disruptors in the OTA landscape.

In conclusion, in the short term, AI agents and OTAs will likely engage in a “cooperative rivalry”—with OTAs overseeing resource allocation and protective systems, while AI agents enhance personalized user experiences. For instance, travelers may book hotel and flight through an OTA while leveraging an AI tool to design unique routes. However, in the long run, if AI agents can bridge the chasms of data monopolization and trust issues, their de-intermediation approach has the capacity to irrevocably alter industry paradigms. The decisive factor will rest on who can fulfill “thousand And one faces” needs at lower costs and establish trust across varying scenarios and channels. It’s certain that the culmination of this battle will not merely be an “AI versus OTA contest,” but a complete reconstruction of the travel service ecosystem.