What will be the impact of Agentic AI in hospitality: significant as the Internet or as insignificant as blockchain and the metaverse?

Specific to the thrust of Max's question, in this context we are referring to search/find from the booking/transaction perspective. In that light...

For me it is the next generation of search.

Depending upon the user, we can now be far more specific with the input to get more targeted output.

We will be able to speak and converse, which is more natural and a next gen over multiple queries or filters.

I do believe that when the buying time comes after the research phase the customer will still 'come' to the point of booking. Whether that be back to the website or because we provided the customer with visual stimuli to confidently convert the business. Seeing is believing if you like.

To reach full potential the agent will have robust set of 'rules' (preferences and scenario) to base search and feedback on, specific to the customer.

In some cases, such as return booking or simple hospitality product needs - room only, dining only; in time some bookings will be fully made by the agent.

Rich product and service offerings still have the edge in bringing the customer to brand.com. There is simply more to explore and engage with and a more robust buying experience.

Significant for input.

The Internet enabled you to find things yourself. Agentic AI will do everything for you. That’s why it’s: Much bigger than blockchain or the metaverse. And closer to how big the Internet was.

“When your grandmother can use a new technology, that’s when you know it will reach widespread adoption.”

I’m paraphrasing this Kevin O’Leary here but the principle is nevertheless highly instructive for why blockchain and metaverse technologies didn’t have prominent adoption in the travel sector. They weren’t all that easy to use; the killer app hasn’t come along yet.

AI search is different. All one needs to do is access the website or download the mobile app, login and ask questions. It’s simple; it’s frictionless.

So, based on this principle of easy adoption, AI will have a tremendous impact on hospitality and travel. Yes, it will be nuanced in terms of the specific directionality (eg. LLMs backing off from in-chat reservations) but the underlying motivation for growth is there.

Agentic AI in Hospitality: Are We Running Before We Can Walk?

The Internet comparison is tempting, but context matters. The Internet succeeded because it required minimal operational readiness from businesses to participate. Agentic AI is fundamentally different — it demands clean data, integrated systems, and real-time connectivity that much of our industry simply doesn't have today.

Here's the uncomfortable truth: a significant proportion of hotels still operate with fragmented PMS, CRS, and CRM environments that don't communicate effectively with each other — let alone with GenAI platforms. Asking those properties to sell directly through ChatGPT or Gemini today is premature optimism.

Google's partnership model with established distribution players reflects this reality. It isn't a vision statement — it's an acknowledgement that the infrastructure gap is real.

From my work supporting technology integration across hospitality businesses, the bottleneck is rarely ambition. It's foundational readiness: data hygiene, system interoperability, and team capability.

Agentic AI will be transformative. But transformation rewards the prepared. Before debating disintermediation, hoteliers must honestly assess whether their tech stack can support the basics of dynamic, real-time distribution.

We cannot skip the walking stage. The industry learned that lesson with blockchain.

If you’re expecting Agentic AI to hit hospitality like the Internet did (fast, messy, and impossible to ignore) - no. If you’re expecting it to quietly fade away like most blockchain pilots and metaverse hotel tours… also no. The reality sits somewhere in the middle. The technology itself is powerful and very real but hospitality is not exactly known for its speed of adoption, especially when it requires coordination across fragmented systems, legacy tech, and multiple stakeholders who all need to agree on “standards” (good luck).

Where this will show up first is corporate travel. That’s where control, compliance, and pre-negotiated rates already exist, making it much easier for companies to deploy their own agents and let them transact within defined rules. Leisure will likely stay in platforms like ChatGPT, because it’s easier for consumers to use what’s already in front of them than to “bring their own agent.” So yes, Agentic AI will reshape our industry but not overnight, and not evenly. It’s less “Internet-level explosion” and more “slow but inevitable shift,” where the real change happens behind the scenes first, long before most travelers realize their trip was booked by two machines politely negotiating on their behalf.

Let's start by saying I don’t agree with the premise: blockchain has found pockets of application, even if marginal, such as tokenized bookings, loyalty programs, identity verification, and baggage tracking. None of these has revolutionized the industry, but “irrelevant” is a strong word.

On the metaverse, if we reduce it to VR social platforms like Horizon Worlds, then I agree with Max. But if we widen the lens to include spatial computing, the industrial metaverse, and the transition beyond a smartphone-centric world, the narrative changes.

Agentic AI belongs to a different category altogether, as it does not create a new channel, but rather operates at the layer where decisions are formed. This is precisely why I do not see it as a force of disintermediation. If anything, it is likely to accelerate this dynamic and reinforce the centrality of OTAs, which already sit at the intersection of inventory, data, and demand. These systems are structurally better positioned to integrate AI at scale and orchestrate agentic interactions.

A real shift may occur on the user side: agentic AI can enhance the individual’s ability to navigate complexity, compare options, and act with greater precision, but I doubt the center of gravity will move toward brand.com.

Remember the facts. “Agentic AI” is just an LLM trying to autonomously orchestrate the actions of other LLMs, including reaching out to standard systems like CRMs, payment gateways etc. All LLMs are still next-best-word, probability-based, correlation engines. They have no internal “world models”. They don’t understand up/down, before/after, time, logic, gravity, distance, and especially numbers. They just mimic based on training data. “Reasoning models” are just rewired LLMs that recycle the same prompt multiple times to try to improve success, but they can’t reliably read an analog clock face. PMS, POS, Spa, IBE, etc. are deterministic systems. Same data in, same answers out. Consistent, accurate, reliable. There is no leeway to almost take the right payment. Almost book the right room in the right hotel etc.

The dotcom era ushered in self-serve for customers to book flights and hotels. Deterministic, global booking engines. Websites democratised distribution. Many call centres weren’t needed and closed. Many high street tour operators closed. Customers did things for themselves and liked it. High-end tour operators thrived as their clientele needed bespoke treatment.

There is some “agentic AI” being linked to call centres now, but they are essentially just duplicating IBEs. Useful for some customers, but not transformational.

Agentic AI is not a tool. Excel is a tool. Similar to the Internet, AI is not a tool, but a transformational technology with huge economic, social, educational, cultural and behavioral implications.

In the same time, Agentic AI is NOT a booking channel. At this point LLMs like ChatGPT and Gemini are mere “travel discovery tools, responsible for 15% of travel discovery. The remaining 85% of discovery? Good old search engines, social media, friends and family recommendations, influencers, etc. 

Travelers are using AI to plan trips, but only 8% are comfortable letting it handle bookings, according to a new survey from Expedia Group. 

Travel discovery is just the beginning of the very convoluted, very complicated travel consumer journey. The Digital Customer Journey is like a meandering river with dozens of digital touch points (Google claims 48 of those). In my view, LLMs will become just one of these touch points. 

After 25 years of trying hard to become a one-stop-travel booking platform (Remember the failed “Book on Google?”), Google has become just one of these touch points. 

Recently, Julie Farago, Google’s vice president for Travel and Local, sought to clarify the company’s intentions. “Google has no intention of becoming an online travel agency,” she said.

Everything has to change, so that everything stays the same. Riffing on Di Lampedusa's famous saying: Agentic AI will have a major impact in hospitality, but that impact will be more focused on the experience of travelers than on the underlying structure of the industry. In other words: most likely, the winners of this new AI era will be the same ones as in the previous era.

Blockchain and the metaverse were technologies in search of a use-case. Agentic AI is a new way of doing things that consumers can't get enough of. There is an unlimited demand for intelligence. However, many people are too quick to arrive to conclusions: the fact that new technology makes something possible does not mean that consumers really want to do that. Disintermediation is therefore overrated: the success of OTAs has never been about technology (booking a hotel is very easy), it's been about marketing, specifically demand aggregation.

In short: as always, the only relevant issue is consumer behavior. If consumers get more value from AI, for real, then every provider will need to offer a top experience that delivers that value.

Agentic AI is reshaping how guests discover and plan travel, but its impact will be evolutionary rather than revolutionary.

AI-driven travel planning has become standard for many consumers. This does not replace existing sales channels; it creates an additional and fast-growing one. For hoteliers, that means more complexity. In the future, hotels must ensure their products, content, and rates are accessible not only to OTAs and search engines, but also to AI-driven platforms.

At the same time, fully autonomous booking decisions will remain limited. Many travellers are still reluctant to hand over complete control of important travel choices, especially for expensive or complex trips. In addition, regulation, particularly in Europe, will significantly slow wider adoption. Questions around data privacy, liability, transparency, and decision-making remain unresolved.

So, the real issue is not whether agentic AI will change hospitality. It will. But it will not replace everything, nor is it the holy grail some claim. Its real significance lies in becoming a new distribution channel that hotels must prepare for.

That preparation starts with access to information, and connectivity. Protocols such as MCP will be essential to connect infrastructure to the AI world and remain visible in the future customer journey.

Agentic AI is the next biggest transformation in our industry, but we only unlock its full potential if it is deeply embedded into operations. We’ve seen technologies like blockchain and the metaverse generate hype without changing how hotels run. The difference with agentic AI is that it can fundamentally change operations, executing tasks and enabling an autonomous operation, not just generate insights.

The real impact comes when AI agents are connected to the underlying systems, with access to real-time data and the ability to follow property-specific SOPs. This is the approach we're taking with Apaleo Copilot, where staff can “talk” to their tech stack using natural language to execute multi-step and multi-system workflows, from changing reservations to updating housekeeping, via a single interface without switching tools.

In this model, AI agents orchestrate workflows across departments and create an autonomous service layer around core systems. However, this depends on an open, API-first infrastructure, where it’s easy to interoperate with standards like the Model Context Protocol. Without such infrastructure, agentic AI will underdeliver. For hotels that are structurally ready, agentic AI will be really transformative.

Blockchain and metaverse were never connected to the industry's core value chain. I wouldn't compare Agentic AI to them. Agentic AI is significantly different. It addresses problems the industry has been trying to solve for decades: distribution cost, operational complexity, labor constraints, and the gap between guest expectation and service delivery.

That puts it closer to the Internet in structural significance. But with a critical distinction. The Internet's impact was broadly accessible. Build a website, attach a booking engine, list on an OTA etc. Agentic AI is infrastructure-dependent. Its impact will be determined not by adoption intent but by data readiness. Hotels with clean, structured, machine-readable inventory exposed through open protocols become discoverable and bookable by AI agents. Hotels without that foundation become invisible to the systems that will increasingly determine traveler decisions.

This is the real divide. Not proponents versus skeptics. It is infrastructure-ready versus infrastructure-absent.

There really have been some heated discussions about the impact of agentic AI on the industry and, assuming we're talking about agentic distribution channels, both sides are correct. Why? Because the real answer is that it will be as impactful as the hotel industry wants it to be.

When the internet emerged, brands held back and waited to see what would happen. Well, OTAs happened, and they dominated online distribution channels to the extent we see today. Brands made some half-hearted efforts to reclaim distribution by restricting loyalty points and stay credits to direct bookings, but they ultimately gave ground to the OTAs, and the OTAs said "thanks so much!"

In agentic channels, the advantages are tilted much more in the brands' favor, but only if brands claim them, and the longer they wait, the harder it will be. Chatbots or agents from companies like OpenAI and Anthropic will try to find the best options for their users, and brands can take steps to ensure that direct bookings are the best matches through dynamic personalization in ways OTAs can't match.

BTW, Google's approach is correct--but only if your aim is to make Google the most money possible!

Thank you Max, this framing makes it easier to answer: The impact will be way more significant than blockchain/metaverse, since it is hard to come up with anything being as insignificant than those two. The internet comparison is more nuanced: as significant as the internet, however not yet. It will likely reach the same level faster, but it still needs time.

Aggregation of options and filtering is always needed, whether done by travel agents, Google search/OTAs plus reviews or an AI layer. Being a (usually) high-involvement decision, most users will doublecheck for some time if the recommended options are a valid selection. If this is true several times, they will skip the time-consuming “search-and-narrow-down” part. It is a bit like GPS navigation – it took some time to build trust in those suggestions, while now trucks get stuck in mud following them blindly.

People probably still will decide on their own between the 3-5 presented options but then let their agent finalize the transaction. And there is a high chance that agent will book directly, either because he is told to do so or direct is the best option anyway. How the agent gets paid remains to be seen.

Indeed, the way we operate and manage hospitality systems is set to undergo a significant transformation with the rise of AI—particularly agentic AI. Adoption is accelerating rapidly. IDC predicts that by 2030, 30% of travel bookings will be executed by AI agents.

Direct bookings through platforms like ChatGPT or models such as Gemini already present clear, low-hanging opportunities. However, the potential applications of AI agents extend far beyond.

While hospitality management systems and ERPs will continue to serve as the system of record, AI agents can support a wide range of operational tasks, i.e., booking, check-in, check-out, housekeeping schedules, incident management and escalations, pricing intelligence, purchasing, and inventory issues and controls.

It is also undeniable that the industry is facing a global staffing shortage. In this context, agentic AI has emerged as a timely solution: augmenting, training, and assisting current teams, and, in some cases, filling roles that employees handle.

That said, they cannot replace the human touch or empathetic decision-making that defines true hospitality. Agentic AI does have the power to catalyse business excellence. It may not completely replace or take over humans, but it will remain significant as the internet, coexisting with human labour.

The internet comparison sets the wrong expectation. Agentic AI will matter, but the timeline is uneven and so is the distribution.

Blockchain in hospitality solved a problem that didn't exist at scale. The metaverse required customers to want something they never asked for. Agentic AI is different: use cases are real, infrastructure exists, demand is forming.

On the operations and marketing side, the case is already solid today. Revenue management, guest communication, agents monitoring campaign performance vs signals like local events, weather and occupancy: these run on structured, owned data and return measurable results. Larger chains are already running this.

On the customer-facing booking side, I'm less convinced. Fragmented inventory and payment logic make hotel transactions harder to automate, but that's solvable. The real issue: a chatbot is a poor purchase surface. Walmart ran 200.000 products through OpenAI's Instant Checkout and reported conversion three times lower than on the website - retail, far simpler than any hotel booking. The interface was the problem. OpenAI stepping back from in-chat transactions proves the same point.

The hotels that see real impact already own their data. Everyone else waits for their vendor to catch up, same as with mobile, structured data, and every shift before this one.

Agentic AI will definitely change things, especially in discovery. But we tend to throw everything into one basket. Travel isn’t one thing. Sometimes people travel because they have to, sometimes because they want to and that changes everything.

For more functional trips, AI will be a game changer. It’ll get you what you want, fast, frictionless, done. That’s powerful.

But for the trips that actually matter, the ones driven by emotion, curiosity, or not even knowing what you’re looking for yet, it’s different. AI can guide, but it can’t fully replace that discovery process.

And one thing won’t change: platform power. Whether it’s Google, OTAs, or GenAI interfaces, where the demand is, you need to be.

So yes, agentic AI will have a real impact. But not evenly. And definitely not in a “one-size-fits-all” way.

Better to be on the train than debating if it will arrive.

In the short term, the impact of agentic AI will be insignificant. Hospitality companies will launch many projects on implementing agents in their operations. Many will fail and will not deliver to the expectations and the promises of software companies. Wrong workflow, wrong booking details and other mistakes will require substantial human involvement. Instead of automation of task, initially the agents may require significant additional human labour. Disillusionment will follow. However, these early attempts will create a steep learning curve for both hospitality and software companies which will help them improve the agents and the processes in which they are implemented. In the long term, the effect of Agentic AI will be profound as it will lead to the automation of many digital processes and the mistakes the agents will make will be minimal. Probably in 10 years, hospitality manages and customers will be able to confidently say "In agents we trust"!

In my view, Agentic AI will be far more impactful than blockchain or the metaverse in hospitality, though not as immediately transformative as the Internet. The Internet reshaped the industry by creating a new layer for distribution, commerce, and information. Agentic AI’s influence will be more gradual, evolving workflow-by-workflow and use case by use case.

The real value goes beyond “AI booking a room.” AI agents will reshape how travelers discover, compare, personalize, and purchase travel. Over time, guests may rely less on traditional search, OTA filters, and brand websites, and more on trusted AI assistants that understand their preferences, loyalty status, budget, and purpose of travel.

That said, full OTA disintermediation is unlikely. Distribution power rarely disappears, it shifts. OTAs, search engines, hotel brands, and AI platforms will compete to become the trusted orchestration layer, with success driven by data, inventory access, loyalty integration, pricing intelligence, and seamless transactions.

For hoteliers, the priority is practical: ensure content is AI-readable, strengthen direct booking, expand trusted distribution channels, and deploy AI internally for revenue management, service, operations, and personalization.

Agentic AI is not hype, but its transformation will be steady, rewarding those who convert new capabilities into measurable results.

Somewhere between the two, but closer to an Internet-level shift than to blockchain or the metaverse.

The Internet reshaped hospitality by changing the fundamentals of information access, planning, and transactions at global scale. Agentic AI has similar potential. It changes who controls planning and decision-making, which can influence economics and will almost certainly change how information is filtered and acted upon.

Today, travelers research, compare, and decide. Tomorrow, they delegate. “Book me somewhere in Lisbon for four nights, good location, no resort fees, flexible cancellation, and use my points if it makes sense.” In that scenario, the agent is not browsing endlessly. It is narrowing the field, screening options, and acting on behalf of the traveler.

Agentic AI compresses the consideration set. Instead of dozens of choices, only a handful may be presented. If a hotel is not clearly understood or trusted by the agent, it may never be considered.

This goes beyond today’s LLM visibility battle. Agentic AI acts. Still, the pace of change should not be overstated. This version of the Internet, enabling this question, took time over multiple iterations before reshaping behavior at scale. Agentic AI will likely follow a similar, uneven path, making it far more consequential than blockchain or the metaverse, but not an overnight transformation.

Agentic AI will matter more to hospitality than the internet did. The internet transformed distribution and moved systems to the cloud — significant, but ultimately confined to a narrow slice of the business. Agentic AI touches everything: distribution, operations, guest experience, and the economics of ownership itself.

The most immediate impact is on people. A newly hired concierge or marketer no longer needs months of shadowing to reach competency. AI provides instant access to unified guest context, multilingual capability, and institutional knowledge that previously lived only in veterans' heads. The learning curve compresses dramatically — and the consistency that once took years to build can now be achieved in weeks.

But the implications run deeper than onboarding. When AI handles the administrative weight of hospitality work — scheduling, reporting, communications, data entry — staff get something genuinely scarce back: time with guests. That reallocation is where service quality is actually won or lost.

For asset owners and operators, this is a structural shift, not an efficiency gain. Labour has always been hospitality's largest cost and its most variable lever. Agentic AI doesn't replace the people who make hospitality worth experiencing — it removes the friction that stops them doing exactly that. The result is an operating model where NOI growth no longer depends on headcount growth, and where great service stops being the exclusive preserve of the best-staffed properties.

The industry has talked about working smarter for decades. The tools to actually do it have arrived.