A new global study from h2c shows AI adoption accelerating rapidly across hotel companies. But after reviewing the research alongside what is happening in consumer AI, our biggest takeaway is that the more important transformation may be happening outside the hotel.
h2c’s new AI Opportunity Study 2026, is one of the more comprehensive looks so far at how hotel chains are actually using artificial intelligence.
The headline is encouraging: AI adoption in hospitality has accelerated significantly over the past year. But reading through the research, another story emerges. Hotels are becoming increasingly comfortable using AI to improve the way they already operate. At the same time, consumer AI is beginning to change the way people search, decide and transact.
Those two developments are happening at very different speeds.
The hotel AI picture in 2026
| What the research found | 2026 |
|---|---|
| Hotel chains already using AI | 91% |
| Using chatbots | 64% |
| Improved operational efficiency from AI | 67% |
| Staff able to focus on higher-value work | 59% |
| Measurable ROI achieved | 13% |
| Company-wide AI strategy led by senior management | 28% |
| AI agents identified as a major area of future innovation | 70% |
| AI booking identified as a major area of future innovation | 70% |
| Systematically monitoring visibility in AI search | 22% |
The numbers show an industry that has moved well beyond simply experimenting with ChatGPT. Adoption has risen from 78% to 91% in a year, while the proportion of companies still exploring their AI approach has fallen sharply.
What hotels are actually doing with AI
The 230 real implementations are perhaps the most useful part of the research. Guest communication and reputation management lead current adoption, followed by revenue management and forecasting, internal productivity, marketing and content, data analysis and reporting, and reservations and booking.
This makes sense. These are areas where AI can already produce practical results without fundamentally changing the hotel business. It can summarize thousands of reviews, answer repetitive guest questions, help revenue managers understand changing demand, create marketing content, interrogate business data, prepare reports and assist reservation teams.
The benefits reflect that. The strongest results are operational efficiency and freeing employees for higher-value work, while measurable financial ROI remains much less common.
So the first wave of hotel AI has largely been about assistance. Make an employee faster. Automate a repetitive process. Analyze more information. Help someone make a better decision.
An agent can potentially research a trip, compare hundreds of options, check loyalty programs, understand preferences, access calendars, choose flights and hotels, enter payment information and complete the transaction. Later, that same agent could continue watching the reservation and take another action if circumstances change.
Consumers are moving from asking AI to trusting it
Hilton CIO Michael Leidinger made a particularly revealing comment at the Destination AI conference this week. He said consumers are already giving personal AI agents their credit card details, something he admitted he had not expected to happen this quickly.
That matters because payment was supposed to be one of the big psychological barriers to agentic commerce.
Leidinger also described what happens when agents move beyond simply finding a hotel. A personal agent could keep shopping after the reservation has been made, find a better price and potentially cancel and rebook. Suddenly the hotel is dealing with software that can shop continuously rather than a traveler who searches a few times before making a decision.
His broader comment was even more consequential for hotel distribution. Large language models, he argued, put OTAs “under real threat” because AI can provide the aggregated view of a fragmented hotel market that historically gave OTAs much of their value.
Whether that ultimately weakens OTAs or simply creates another generation of intermediaries remains an open question. Leidinger himself expects personal agents eventually to become commissionable channels. What matters today is that the interface between the traveler and the hotel industry is beginning to change.
And suddenly everyone is building that interface
Look at what has happened around travel AI in a remarkably short period.
Booking Holdings has backed Lola. Expedia acquired Layla in July. Meta has introduced Muse, a personal agent designed to operate across websites and services. OpenAI has now announced Dots, persistent agents intended to keep working on behalf of users over time. Google is bringing agentic hotel booking into AI experiences, while hotel groups themselves are building connections into the major AI platforms.
This is moving beyond the familiar idea of replacing Google’s search box with a conversational search box. The potentially much bigger change comes when the consumer stops conducting every individual search.
The agent does it for them.
This is where the h2c research becomes particularly interesting
Hotels clearly understand that something is happening. AI agents and AI booking are the two most frequently identified areas of expected innovation in the study.
But preparedness tells another story.
While most hotel chains hope AI-driven booking channels to increase direct bookings over the next couple of years, only a minority systematically monitor how their hotels currently appear in AI-generated results. A third have yet to implement specific measures aimed at improving their visibility and bookability through AI channels.
And perhaps the most important constraint is underneath all of this.
As h2c Director Christin Haensel puts it:
“Hotels do not simply need more AI features. They need connected systems, accessible and well-governed data, and interoperable platforms that enable AI to work reliably across commercial, operational, and guest-facing workflows.”
That may be the practical takeaway for hoteliers.
The question is increasingly less about whether your PMS has an AI button or whether your marketing department uses ChatGPT. The bigger question is whether an intelligent system can understand your hotel and actually do business with it.
Can it find accurate room information? Can it understand amenities and policies? Can it access live rates and availability? Can it compare room types? Can it understand loyalty benefits? Can it make a reservation? Can another authorized system change or cancel that reservation?
If the consumer’s agent becomes the interface, these become distribution questions as much as technology questions.
Two AI adoption curves are about to meet
The h2c study should ultimately be read as good news for hospitality. An industry often accused of moving slowly with technology has gone from 78% AI adoption to 91% in just one year. AI is already producing practical improvements across operations, revenue management, guest communication, marketing and analytics.
But there is another adoption curve worth watching.
Consumers went from experimenting with ChatGPT, to using AI for research, to asking it to plan trips, to connecting it to personal information, and now, in some cases, trusting agents with payment credentials and actions.
That progression has happened extraordinarily quickly.
Hotels therefore need to watch more than their own AI adoption. They need to watch their customers’ AI adoption.
Because the next major change in hotel technology may not start inside the hotel at all. It may start with a guest saying to their personal AI:
“Find me somewhere to stay, make sure it’s good, and book it.”
And then letting the agent take it from there.
