Simon Lehmann, CEO at AJL Atelier: “We’re Moving from Platform-Centric to Access-Centric Distribution”
For decades, lodging distribution has been shaped by online travel agencies (OTAs). By aggregating both supply and demand, they have become a central gateway between hospitality providers and travelers.
Now, conversational AI and agent-based interfaces may begin to change that dynamic. Emerging frameworks — such as the Model Context Protocol (MCP) and the Universal Commerce Protocol (UCP) — could make it easier for AI systems to access and use content in real time.
Simon Lehmann, Co-Founder and CEO of AJL Atelier, with over 35 years of experience in hospitality and more than two decades in the short-term rental (STR) sector, shares his perspective on how this transition could gradually transform distribution.
How AI agents challenge OTA dominance in hotel and STR distribution
At a high level, I believe we are not just looking at another wave of technology, but at a structural shift in how demand flows in hospitality. For years, distribution in hospitality has been controlled by a small number of large players, and direct bookings have remained difficult to scale.
What is changing now is the interface itself. We are moving from platform-driven discovery to conversational and agent-driven discovery. Instead of users navigating marketplaces, they increasingly describe intent, context, and preferences, and AI systems translate that into options. This may seem like a UX change on the surface, but it has deep structural consequences.
One of the most important implications is that aggregation, a core advantage of OTAs, is starting to weaken. Historically, OTAs won because they controlled both demand and supply aggregation. But if AI assistants sit upstream and curate options from across the entire web, then discovery is no longer owned by a single platform. The user may spend more time in an AI interface than on any individual OTA, which fundamentally challenges the traditional model.
This shift creates a potential opening for operators in the short-term rental market, where supply is highly fragmented across thousands of hosts. In the traditional model, visibility comes at a high cost — especially in search. AI-driven discovery works differently. It breaks down user intent into granular queries and evaluates a much broader set of sources, effectively widening the aperture and allowing smaller players to surface alongside large platforms — provided their content, positioning, and underlying data are structured correctly.
What the future holds for OTAs in an AI-driven ecosystem
I do not believe the agentic AI leads to the disappearance of OTAs. 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. The future is likely a hybrid model in which direct and third-party distribution coexist, with a healthier balance for STR operators.
There are also a few risks and open questions that should not be underestimated. One is the emergence of new gatekeepers. Control may shift from traditional software vendors and OTAs to AI platforms and orchestration layers. Another is around trust, governance, and liability when agents start acting on behalf of users. And finally, standardization through protocols like UCP and MCP could significantly reduce switching costs, increasing competition and accelerating commoditization across the stack.
Overall, I would frame this shift less as a technology story and more as a redistribution of control. We are moving from a platform-centric model to an access-centric model. The winners will be those who control their data, expose it effectively, and position themselves clearly enough to be understood and selected by AI systems.
How AI could change direct booking dynamics
Historically, the industry has struggled with direct bookings because it is not a single problem. It is a combination of acquisition, pricing, trust, conversion, and retention. As discussed in my podcast, this “combination lock” has been very difficult to solve consistently, especially for independent operators.
AI changes this in two ways. First, it improves the ability of STR operators to execute across all these dimensions more efficiently. Second, and more importantly, it changes guest behavior. If AI assistants begin to understand that direct bookings often provide better value, that knowledge can be embedded into the decision-making process itself. In other words, awareness of direct booking advantages may no longer need to be created through marketing, but could be inferred and recommended by the assistant.
Why protocols like MCP and UCP matter
These types of protocols are not just technical standards. They enable a new access layer. We are moving from systems of record to systems of access. In that world, dashboards and interfaces lose importance. What matters is whether your data can be accessed programmatically, understood by machines, and used in real time. This is why I believe data accessibility will become more important than data visualization. The industry has spent years optimizing how humans look at data. The next phase is about how machines use it.
Where hospitality stands in MCP/UCP implementation
In terms of real-world examples, I would say we are still early. There are experiments happening, especially on the OTA side and among some AI-first startups, but we are not yet at a stage where MCP or UCP are broadly implemented as industry standards. What we are seeing instead are early integrations where platforms expose inventory and pricing to AI systems in more dynamic ways. So at this stage, it is more forward-looking with pockets of experimentation rather than full-scale adoption.
What will drive MCP/UCP adoption
My view is that it will not be the hotels or STR operators themselves. Adoption will be driven by infrastructure players. PMS providers, channel managers, and large distribution platforms are in the strongest position because they control access to inventory and data flows.
Hotels and operators will follow once these capabilities are embedded into the systems they already use. This is consistent with how most technological shifts have happened in hospitality. Suppliers rarely adopt protocols directly; they adopt products that abstract them.
At the same time, AI platforms themselves will be a major force. If companies like OpenAI, Google, or others create strong incentives or standards for accessing inventory, that will accelerate adoption across the ecosystem.
How AI agents could reshape post-booking communication
Today, communication in hospitality is fragmented across email, messaging platforms, PMS systems, and OTAs. In an agent-driven environment, communication could become more centralized and continuous
The interesting question is who owns that relationship. If the booking originates through an AI assistant, there is a possibility that part of the communication layer also sits there, at least initially. However, operational communication will still need to be handled by the hotel/STR operator or their systems.
So in practice, I would expect a hybrid model. AI agents may handle pre-booking and parts of post-booking communication such as confirmations, FAQs, and basic requests, while more complex or sensitive interactions are managed by suppliers or their tools. Over time, this could evolve into a shared communication layer where responsibility is distributed but needs to be clearly defined.
How hotels and STRs can prepare for agentic AI
Most hotels and STR operators should not start with the protocols themselves. They should start with their data layer.
If a hotel group or operator wants to experiment in this space today, the first step is ensuring that their data is structured, accessible, and consistent across systems. That includes availability, pricing, content, guest data, and operational data. Without that foundation, protocols like MCP or UCP do not create value on their own.
From a technical-readiness perspective, this means having clean APIs, well-structured content, and, ideally, a centralized data layer rather than fragmented systems. From an organizational perspective, it requires alignment between commercial, revenue, and tech teams. In many organizations, those are still siloed, which becomes a major blocker.
When hospitality will be ready for full-agent booking
I agree with the skepticism in the short term. The industry is not ready for full agent-driven booking today.
There are a few key blockers. The first is trust. Booking travel is a high-consideration decision, especially for leisure. Users are not yet comfortable delegating that fully to an AI.
The second is verification. As discussed in my podcast, many travel decisions are subjective and difficult to validate upfront, which makes full automation harder.
The third is liability and accountability. Someone needs to be responsible if something goes wrong, and that is not yet clearly defined in an agentic model.
The fourth is user experience. Current AI interfaces are improving quickly, but they are still not at the level where they can consistently outperform existing booking flows.
That said, I do believe we will see booking through AI interfaces over time, but it will happen gradually and in stages. It will likely start with lower-risk, more standardized transactions such as business travel or repeat bookings, before moving into more complex leisure scenarios.
So in summary, discovery will change first and most significantly. Booking will follow, but more slowly. And full end-to-end agentic booking will only happen once trust, UX, and accountability frameworks are sufficiently mature.