Who Will Control Lodging Distribution in the Era of Agentic AI? 7 Questions to Hospitality Experts

Over the past two decades, lodging bookings have been largely dominated by Online Travel Agencies (OTAs), which control over 60 percent of distribution for independent hotels.

But that position is now being challenged, as generative AI models are poised to reshape how travelers discover and buy stays. AltexSoft spoke with industry experts to explore whether this shift could rebalance the market, who will ultimately control different parts of the end-to-end booking flow, and other critical questions surrounding the future of hotel distribution.

Where is GenAI already having the biggest impact in hospitality?

The traveler journey includes several key stages

• discovery, when people search and compare options;
• booking, when reservations and payments happen; and
• post-booking, which covers everything after the reservation — itinerary changes, guest communication, customer support, and handling disruptions.

Today, GenAI adoption is moving the fastest at the discovery stage. As Simon Lehmann, Co-Founder and CEO of STR consultancy AJL Atelier, explains, “instead of users navigating marketplaces, they increasingly describe intent, context, and preferences, and AI systems translate that into options.”

Traditionally, the discovery stage has been largely controlled by OTAs, which aggregate both supply and demand bringing multiple lodging options into a single platform where hotels and short-term rental (STR) operators compete for visibility. “The reality is simple: Booking.com and Expedia offer centralized inventory, delivering what users actually want — choice,” says Ira Vouk, Technology Strategist and Founder of the AI Hospitality Alliance. “People don’t want to browse hotels one at a time; they want a database they can compare, filter, and sort.”

Not only inventory but also ranking algorithms and much of the communication are managed within the OTA platform, allowing it to influence travelers' decision-making.

Now, that dynamic is starting to change. AI agents can potentially sit upstream of traditional travel platforms, surfacing lodging options from across the web. “The user may spend more time in an AI interface than on any individual OTA, which fundamentally challenges the traditional model,” Simon says.

For smaller players, this could create new opportunities to gain visibility without relying as heavily on expensive search rankings or OTA placement. Instead of favoring only the biggest brands or highest bidders, AI-driven discovery “breaks down user intent into granular queries and evaluates a much broader set of sources,” Simon explains. This could give niche operators a better chance to appear alongside major platforms — provided their content and inventory data are structured in a machine-readable, AI-accessible format.

Are LLM providers ready to go beyond discovery?

In agentic-assisted discovery, which is the most common setup today, a conversational AI interface helps travelers search, compare, and select options, then redirect users to the OTA or hotel website to complete the purchase. But will AI agents move beyond this? And if so, who becomes the merchant of record? In other words, who takes responsibility for payments, refunds, chargebacks, fraud management, and regulatory compliance?

Historically, hotels handled booking fulfillment and payments, while OTAs largely owned the booking flow and customer interface through the agency model, passing transaction details to suppliers. Expedia later shifted this dynamic by expanding into the merchant model and taking greater ownership over cash flows.

Currently, OTAs may combine both models depending on the supplier agreement. In direct bookings, however, the hotel remains the merchant of record. But what happens when a new intermediary — the GenAI interface — enters the process?

Let’s look into possible scenarios.

Agent-moderated commerce enables travelers to search and complete bookings without leaving the conversational interface. One-click reservations and integrated payments create a much more seamless experience for users. But in reality the merchant of record can be the supplier or an intermediary behind the scenes.

One example is Perplexity AI, which enables hotel discovery and instant checkout with credit cards, PayPal, or Venmo directly within the chatbot interface. However, the transactions themselves are handled by Selfbook, a hotel commerce platform that powers the underlying booking and payment infrastructure.

Another LLM provider — OpenAI — ultimately removed native checkout functionality from ChatGPT. While the technical infrastructure existed, other challenges quickly emerged: limited consumer trust, the complexity of live data synchronization, and regulatory compliance requirements. And travel only amplifies this complexity.

“I think many underestimated how complex and fragmented travel really is,” says Paul van Alfen, Managing Director at travel payment consultancy Up in the Air. “It’s not a plug-and-play industry. Buying travel is very different from reordering a simple commodity like detergent. It involves disruptions, post-booking servicing, refunds, risk management, live inventory synchronization, and data security.”

While it may be technologically feasible for LLM providers to own the booking stage and transactions in the AI-moderated commerce scenario, many experts remain skeptical that they will fully move in that direction anytime soon.

“I don’t see that happening right now,” Paul van Alfen says. “To do it right, you have to fully commit — invest in systems, resources, compliance, regulation, and operational infrastructure. You don’t become Booking.com overnight.”

End-to-end agentic booking is the most ambitious and futuristic scenario involving autonomous agent-to-agent transactions. In this setup, ChatGPT, Perplexity AI or any other GenAI platform interacts directly with the hotel’s or OTA’s agent and completes the booking on the traveler’s behalf  — without manual involvement.

“It raises the biggest questions about identity, trust, payments, who is doing what, who owns the transaction, and who is the merchant of record. This model is not yet operating, largely because we haven’t solved its core issues yet,” Ira Vouk explains. “Full end-to-end agentic booking will only happen once trust, UX, and accountability frameworks are sufficiently mature,” Simon Lehman confirms.

No matter the agentic scenario, the ideal situation, according to Ira, is “for the hotel to always remain the merchant of record.” The hospitality business stays responsible for customer communication and support, while any intermediaries between supply and demand simply facilitate the transaction.

Will AI agents kill OTAs?

While AI interfaces reshape the discovery process on the surface, behind the scenes they still need access to multiple sources of accommodation supply. And this is exactly where OTAs continue to hold a major advantage. “More likely, we will see a rebalancing. OTAs may lose their dominance as the primary interface, but they will remain important as inventory providers, trust layers, and transactional intermediaries,” Simon Lehman says.

Their strong positioning is only confirmed by the fact that Booking.com and Expedia became the first travel brands to launch their apps in ChatGPT.  For AI platforms, connecting to a handful of large players is significantly easier than integrating directly with hundreds of thousands of individual hotels, each operating different property management systems (PMSs), content formats, and connectivity standards.

If not OTAs, who will make hotels visible in AI-driven travel search?

While OTAs are unlikely to disappear, a new type of AI-native aggregator is beginning to emerge alongside them, designed to give hotels access to AI-driven discovery platforms. .

“An AI platform connects to a single aggregator that consolidates direct booking inventory, then redirects demand to hotel websites for a modest commission,” Ira Vouk describes the model. “The key difference is that hotels remain the merchant of record and keep all customer data — communication, profiles, even just the email address — which OTAs don’t provide today.”

Tomotaka Hirabayashi, Data Driven Re-Design Strategy Lead Partner at EY Japan and  Committee Member of Japan Tourism Agency, calls these new players “intermediaries 3.0”: “It’s not OTA 2.0, because the new model is fundamentally different from the OTA paradigm. It’s a more C2C-like booking flow that deeply involves suppliers in the process.”

Experts point to Google as an example of this aggregation model. Its AI Mode conversational interface supports hotel discovery by leveraging the connectivity infrastructure originally built for Google Hotels and Google Hotel Ads. Google acts as a discovery and routing layer, directing travelers to supplier or partner websites to complete the booking.

However, the ecosystem is not purely hotel-centric and still relies heavily on large OTAs such as Booking.com and Expedia Group. “ What the industry needs is a similar approach for direct bookings. If that happens, it could improve profitability for hotels and deliver a better experience for travelers,” Ira Vouk says.

A smaller player in this space is DirectBooker — an AI-native aggregator that connects independent hotels and hotel brands to major AI platforms for fees far below typical OTA commissions. Each recommendation includes a direct booking link to the hotel’s own website, with trip details such as dates, occupancy, and preferences already pre-filled. If a property operates its own AI assistant, DirectBooker can also route travelers into that conversational booking experience to complete the reservation.

Will protocols like MCP and UCP become the new standard in hospitality?

As AI systems become more involved in travel discovery and transactions, the industry is starting to explore broader AI communication protocols that could help standardize how AI agents interact with hospitality systems.

One of the emerging approaches is Model Context Protocol (MCP), a set of rules that allows AI agents to communicate directly with business systems without requiring custom integrations for every bot. Some travel companies are beginning to test how it could support AI-driven discovery and distribution.

For instance, Aven — formerly part of Sabre Hospitality—is embedding MCP into SynXis, its central reservation system (CRS) serving more than 35,000 hotels. If rolled out as planned, these properties could connect to AI-driven discovery channels directly through the platform while retaining control over pricing, loyalty programs, and discount rules.

While MCP supports the discovery layer, another technology—Google’s Universal Commerce Protocol (UCP)—is designed to coordinate interactions between AI agents, payment providers, merchants, and business systems, enabling authorization, checkout, order management, and other transactional workflows. In theory, it could power both booking and post-booking processes in hospitality — handling payments, confirmations, modifications, and data exchange directly inside AI conversations. However, adoption remains very limited for now, and real-world implementations are still at an early stage.

According to Ira Vouk, in an ideal scenario, “we’d have an MCP layer for discovery and a UCP layer for transactions, both tailored to hotels. But in reality, not all hoteliers will implement them, and some will build their own solutions anyway.”

Hospitality has spent decades trying to standardize — with limited success. As Ira Vouk explains, “No single standard can address the variety of use cases that exist in our industry.”

Out of roughly 1.3 million hotels worldwide, more than a million are independent, each operating differently. On top of that, the industry relies on a fragmented technology landscape serving different business models, segments, service levels, and property sizes.

“So any attempt to impose a one-size-fits-all standard is unlikely to gain traction,” she concludes. “. For example, Google hasn't adopted MCP. Not yet.”

Will travelers finally trust their money to AI agents?

As of early 2026, around 37 percent of travelers use large language models (LLMs) integrated into travel websites to plan and book their trips.

However, readiness to plan — and even reserve — through AI is not the same as readiness to pay within an AI interface. As Simon Lehmann notes, trust remains a key barrier: “Booking travel is a high-consideration decision, especially for leisure. Users are not yet comfortable delegating that fully to an AI.”

Recent Skift research suggests that only about 2 percent of travelers would allow an AI agent to complete a purchase on their behalf, while an Expedia Group survey puts the figure at 8 percent — still a small minority.

That said, this is likely to be a temporary barrier. As Ira Vouk explains, “The transaction layer isn’t new — payments are handled by existing systems like Stripe. What will change is how seamless the experience becomes — keeping everything inside one interface. It’s still messy today, but that will improve.”

The timeline may resemble previous shifts in consumer behavior. Mobile commerce, for example, was set for mass adoption with the iPhone in 2007 and Android in 2008, but only became a significant force in retail around 2016 — nearly a decade later. AI-driven transactions may follow a similar adoption curve. According to International Data Corporation (IDC), up to 30 percent of bookings could be autonomously completed by AI agents by 2030.

How should hotels get ready for agentic distribution?

For hotels, the transition to agentic discovery — and, in the longer term, fully autonomous booking — doesn’t start with protocols. It starts much earlier: with the foundational preparation work.

Prepare data for GenAI. The first and most critical step is the data layer. If hotel data is scattered across platforms such as a property management system (PMS), central reservation system (CRS), channel manager, and booking engine, AI agents would not provide any real value ending up operating on incomplete or inconsistent information. That can lead to inaccurate recommendations, outdated prices, failed bookings, duplicated tasks, or poor guest experiences.

Simon Lehmann emphasizes that data must be “structured, accessible, and consistent across systems—including availability, pricing, content, guest data, and operational data.” He adds that, in an ideal scenario, there should be “a centralized data layer, rather than fragmented systems.”

Reach out to your current distribution providers. Hotels and STR operators rarely build infrastructure from scratch—they rely on technology vendors. Ira Vouk recommends “asking your distribution providers about their plans for AI discoverability and MCP.”

This simple question quickly reveals the real situation. If a provider lacks a clear strategy for AI-driven distribution, the hotel risks falling behind the channels where demand is already forming. As Vouk puts it, “if they don’t have an answer, it may be time to reconsider the provider.”