The Next Frontier of Hospitality AI: From Intelligent Assistants to Autonomous Agents
What h2c's AI Opportunity Study 2026 reveals about how hotel chains plan to use AI agents, and what still stands in the way.
AI agents have become the technology hotel chains are watching most closely. In h2c's AI Opportunity Study 2026, 70% of participating hotel chains name AI agents as a top area of expected innovation for the next two years. Bookings via AI platforms such as ChatGPT and Gemini also reached 70%, tying with AI agents as the top innovation area. The study draws on 122 responses from 113 hotel chains, which together operate more than 8,200 properties and about 1.3 million rooms.
This article looks at what the data says about the move from AI assistants to autonomous agents: how fast it is happening, where hotel chains expect agents to add value, and why most are not yet ready to let them act on their own.
Why are hotel chains turning to AI agents now?
Because AI is already part of daily hotel operations, and the next step is AI that acts rather than just answers. 91% of hotel chains in h2c's study already use AI, and AI agents are now the fastest-rising innovation priority, up 22 percentage points on 2025.
Source: h2c AI Opportunity Study 2026. AI agents and bookings via AI platforms tie as the top innovation areas.
Adoption is moving just as fast. The share of hotel chains in the study using AI agents today rose from 11% in 2025 to 24% in 2026, and a further 44% plan to adopt them. In one year, agents have gone from a talking point to something a quarter of the chains surveyed already run.
Source: h2c AI Opportunity Study 2026. AI agent use more than doubled between the 2025 and 2026 surveys.
Results are still catching up with expectations. Only 13% of hotel chains in the study report measurable ROI from AI so far. For most groups, the question is no longer whether to use AI but how to make it work across the business. Our guide to AI in the hospitality industry covers the wider landscape; this article stays with agents.
What is the difference between an AI assistant and an AI agent?
An AI assistant reacts: it answers a guest or staff question, then waits for a person to act. An AI agent is proactive: given a goal, it works across hotel systems to complete the task and reports back. h2c describes agents as AI capable of completing tasks and making decisions autonomously.
Take a guest who asks for a late check-out. A chatbot can quote the policy, but someone at the front desk still has to check occupancy, approve the request and add the charge. An agent can check availability, offer a paid late check-out, update the booking and the bill, and tell the team what it did. Same request, very different amount of work left for staff.
Illustrative example. Usage shares: h2c AI Opportunity Study 2026.
| AI assistant | AI agent | |
|---|---|---|
| How it works | Reacts to a question or command | Works toward a goal, step by step |
| Who completes the task | A person, after the assistant answers | The agent, within limits the hotel sets |
| Systems involved | Usually one channel, such as chat or voice | Several systems, such as the PMS, CRM and payments |
| Late check-out example | Quotes the policy | Checks occupancy, offers a paid late check-out, updates the booking |
| Use among hotel chains in h2c's study | 58% today | 24% today |
That difference explains the adoption gap. Assistants such as voice-based AI or chat tools on omni-channel guest communication platforms are usually the starting point, and agents can extend those same channels. For a broader introduction, see our earlier piece on how AI agents work in hospitality, or IBM's general overview of AI agents.
Where can AI agents deliver value in hotels?
Mostly behind the scenes, for now. Among hotel chains using or planning to use AI agents, 70% apply them, or plan to, to staff-facing work, compared with 55% for guest-facing work. Respondents could choose both, so the two figures overlap rather than add up to 100%.
The split changes by region. In Asia Pacific (75% staff-facing, 33% guest-facing) and the Americas (75% and 38%), agents are mainly an internal tool. Europe is more balanced at 67% and 62%. The Middle East and Africa is the only region where guest-facing use matches staff-facing use, at 80% for both.
Source: h2c AI Opportunity Study 2026. Among chains using or planning to use AI agents; multiple answers allowed.
When h2c asked where agents should be allowed to act on their own, the answers pointed to revenue and reporting. The top guest-facing value area is upselling and cross-selling (64%). The top staff-facing value area is reporting and insights (61%).
Source: h2c AI Opportunity Study 2026. Results come from two separate questions and are shown separately.
Both are areas where hotels already run dedicated tools, such as hotel upsell software and AI hotel operations platforms, so agents have clear workflows to plug into. Reporting is also where the business intelligence decisions hotels must get right in 2026 start to pay off: an agent that pulls the morning numbers together is only as useful as the data behind it.
What is stopping hotel chains from letting agents act on their own?
Trust and infrastructure. On guest-facing decisions, five concerns are each cited by more than half of responding hotel chains: unclear liability if the AI makes a mistake (56%), data privacy and security (54%), guests' willingness to share personal data (53%), accuracy, reliability and bias in AI outputs (52%), and the limited maturity of current AI solutions (52%).
Source: h2c AI Opportunity Study 2026. Concerns about agentic AI making autonomous decisions in guest-facing interactions; multiple answers allowed.
No single concern dominates, so fixing one still leaves four open. A clear liability policy does not answer the privacy question, and better data does not settle who is accountable when an agent gets a booking wrong.
The bigger obstacle may be the plumbing. Agents need to share information with each other and be coordinated, yet only 9% of responding hotel chains have an AI agent communication layer, and 8% an orchestration layer. Set against the 70% who see agents as a top innovation area, that is a wide gap between ambition and readiness.
Source: h2c AI Opportunity Study 2026. Figures come from different questions.
"AI agents are where hotel chains see the next step, but most are not ready to let them act on their own yet. Only a small minority has the communication and orchestration layers that agents need. Before scaling autonomous workflows, hotels should make sure their systems are connected, their data is accessible and well governed, and accountability is clear."
Christin Haensel, Director, h2c
Skills are part of the picture too: hotel chains in the study rate their internal AI knowledge at just 3.4 out of 10. We made a similar point in our agentic AI reality check: hospitality's systems problem is bigger than the chatbot. It is also why integration is becoming more important than features when hotels choose technology.
How can hotel chains prepare for autonomous workflows?
Start with the three conditions h2c points to: connected systems, accessible and well-governed data, and clear accountability. Each can be checked before a single agent goes live, and each gives buyers a concrete question to put to vendors.
Agents sit on top of the data layer and the hotel systems beneath it.
Connect the systems agents will act on. An agent that cannot read occupancy or write to the bill cannot finish a task. Ask vendors: which of our systems can your agent read from and write to today, and through which integration? Our connectivity and integration platforms category is a useful place to compare options.
Make data accessible and governed. Agents act on guest profiles, rates and inventory, so that data has to be complete and permission-controlled. Ask vendors: what data does the agent need, where does it come from, and who can see what it does with it? Many groups start by consolidating guest data in a customer data platform.
Decide who is accountable. With unclear liability the top concern (56%), set the limits before launch: which actions an agent can take alone, which need approval, and how every action is logged. Ask vendors: can we set approval rules per action, and can we audit every decision?
A narrow first project helps. Reporting is the top staff-facing value area and carries less guest-facing risk, so it is a sensible place to prove an agent works before it talks to guests. When you reach the shortlist stage, our practical evaluation framework helps compare vendors on integration and governance, not just features. You can also browse AI agent solutions for hotels on ExploreTECH.
Explore the full study
This article covers one part of the research. h2c's AI Opportunity Study 2026 also looks at AI adoption, implementation models, data readiness, guest personalization and AI search visibility, and it analyzes 230 AI use cases that hotel chains have already implemented. Read the full findings on h2c.de.
Frequently asked questions
What are AI agents in hotels?
AI agents are proactive AI systems that complete tasks and make decisions on their own, within limits the hotel sets. Unlike chatbots, which answer and wait, an agent can work across hotel systems to finish a task, such as offering a paid late check-out and updating the booking.
How many hotel chains use AI agents?
In h2c's AI Opportunity Study 2026, 24% of participating hotel chains use AI agents today, up from 11% in 2025. Another 44% plan to adopt them.
What is the difference between staff-facing and guest-facing AI agents?
Staff-facing agents support internal work such as reporting, finance and administration, and team coordination. Guest-facing agents deal with guests directly, for example by handling booking requests, answering inquiries or making upsell offers. Among chains in h2c's study that use or plan AI agents, 70% apply them to staff-facing work and 55% to guest-facing work.
What do hotels need before using AI agents?
Connected systems the agent can read from and act on, accessible and well-governed data, and clear rules on accountability. Only 9% of hotel chains responding to h2c's study have an AI agent communication layer and 8% an orchestration layer, so for most groups the groundwork comes first.