Hotelier's Digest #14: What Should AI Handle at Your Hotel?
The hotel industry is having two conversations that seem unrelated.
One is about creating more distinctive hotels: properties with a clear identity, stronger connections to place, and experiences that guests cannot find everywhere else.
The other is about AI and how much of the work inside a hotel can now be automated.
I think they are increasingly the same conversation.
The more capable hotel technology becomes, the more important it is to decide what should be standardized and what still requires human judgment. An AI agent can answer a question, recommend a room, and increasingly complete the booking. But technology cannot decide what kind of hotel you want to operate or what should make a guest choose you in the first place.
This week, I have been thinking about that tension through three different perspectives: a conversation with lifestyle hotel entrepreneur Eric Jafari, a demonstration of an AI agent built specifically for hotels, and a practical playbook for deciding where AI belongs in hotel operations.
Together, they point to a more useful question than simply asking how hotels should use AI:
What should technology handle so your people can spend more time protecting what makes the property worth choosing?
Featured Analysis: Lifestyle Needs an Operating Idea
"The magic happens when we make mistakes. It's when we experiment, when we're willing to take risks."
— Eric Jafari
In my recent article, The Rise of Lifestyle Hotels: What It Means for Independent Hotels, I argued that a lifestyle hotel is not simply an aesthetic. It is a coherent idea expressed through design, service, shared spaces, operations, and a clearly defined target guest.
That argument is put under useful pressure in Matt Welle's conversation with Eric Jafari. Jafari, who co-founded Locke, edyn, and Birch and is now building AENDRE, argues that the lifestyle hotel category has been weakened by scale, imitation, and safer decision-making. He describes an earlier version of the lifestyle hotel as livelier and more focused on bringing guests together. What emerged as a distinct category from the 1980s through the 2000s was gradually replicated or acquired by larger brands, transforming, in his comparison, from the neighbourhood indie café into a Starbucks.
The name "AENDRE" derives from Norwegian and Danish, meaning "to change" or "to alter." It reflects our commitment to transforming spaces and creating something extraordinary from the ordinary.
This does not contradict the opportunity for independent hotels. It clarifies it.
Independent properties often have what larger groups try to manufacture: a genuine connection to place, local guests using public spaces, and owners or operators who know why the property exists. The risk is leaving those strengths informal. If the experience depends on one founder being on site, one employee remembering a repeat guest, or one strong event carrying the atmosphere for a month, the concept is distinctive but fragile.
The work is to turn taste and intent into an operating model without sanding away the character. Define who the property is for. Give public spaces a purpose. Decide which service choices matter enough to protect. Build staff authority around those choices. Then align photography, room descriptions, rate position, distribution, arrival, and follow-up communication with the same idea.
A lifestyle strategy should not make a hotel feel more fashionable. It should make the hotel easier to understand and more reliable in delivering what makes it worth choosing.
Revenue and Commercial Strategy: An Agent That Can Complete the Booking
I met with Jeong Pyon, co-founder of Mizumi, for a product demonstration. The platform is built around hotel workflows rather than a generic chat interface.
During the demo, the agent answered guest questions, helped select a room, and completed a booking through an integration with Mews. It also recommended nearby restaurants and continued conversations across the hotel website, WhatsApp, Facebook, and Instagram in multiple languages. Mizumi is also developing a call agent to handle similar interactions by phone.
That matters because hotel inquiries do not arrive according to office hours. A useful agent can shorten the gap between a guest asking a question and completing a reservation. It can also keep more of the booking journey within channels the property can see and manage.
The integration is where much of the value depends. An agent can only work with the information and systems behind it. Availability, rates, room descriptions, policies, and booking rules all need to be accurate if the agent is expected to answer questions and complete reservations correctly.
I saw the importance of that foundation firsthand while documenting a hotel’s transition from roomMaster to Mews in Moving from roomMaster to Mews: What Hotel Data You Need to Extract. Clean room and rate mappings, clear policies, and reliable data were essential to the migration. The same principle applies here. Before evaluating what an agent can say or do, make sure the PMS gives it accurate information to work with.
Mizumi is worth watching because it has been designed by experienced hoteliers. The demo reflected an understanding of reservations, guest service, destination knowledge, and the interruptions that shape a real front desk. The next test is whether that understanding holds up consistently across live properties, unusual requests, and situations that need to be handed to staff.
Operations and Leadership Insight: Start With the Work, Not the Tool
Piers Thackray's AI Playbook for Hotels is a practical 29-page guide for owners, general managers, and operations leaders. Its strongest contribution is the order in which it approaches adoption. It asks teams to identify recurring work, compare impact with effort, and choose contained projects that can remove busywork without weakening the guest experience or disrupting current systems.
Th1, Inc.
That is a better starting point than asking where a hotel can “use AI.” Start with a specific workflow: unanswered after-hours inquiries, repetitive pre-arrival questions, daily report assembly, restaurant recommendations, or follow-up messages. Record how often the work occurs, who handles it now, what information they need, and what happens when the answer is wrong.
Then define the boundary. A question about breakfast hours carries far less risk than a cancellation dispute, payment problem, accessibility request, or service recovery decision. The important distinction is knowing which tasks can be completed automatically, which require approval, and which should move directly to a person.
For an independent property, a useful first pilot should be narrow enough to supervise and clear enough to evaluate. Track response time, booking conversion, staff time, escalation rates, errors, and guest feedback. If the pilot does not reduce workload, improve the guest experience, or help the team make better decisions, adding more automated workflows is unlikely to fix the underlying problem.
Worth Watching: Hotel AI Moves From Advice to Action
"This quarter, it does the thing itself and leaves the judgment calls for a person."
— Jordan Hollander, Hotel Tech Report
Published July 14, Hotel Tech Report’s Q2 2026 AI Trends and Tactics report reviews 40 products and points to a broader shift from AI that recommends actions to AI that can carry them out across revenue management, sales inquiries, direct booking, guest communication, operations, and finance.
The report is useful as a market signal, but the more important issue is control. As systems gain the ability to change a rate, answer a guest, follow up on a lead, or complete a task, properties need clear guardrails, defined ownership, and a reliable path for exceptions. Automation should reduce repetitive work and unnecessary decisions. It should not make it harder to understand why an action occurred, what information triggered it, or who is responsible for reviewing the result.
This is where the ideas in the AI Playbook connect with the Mizumi demo. An agent becomes more useful when it has reliable hotel data, clear boundaries, visible activity logs, and an easy way to hand a conversation or decision to a person. Without those foundations, a system may complete more tasks while making the property harder to manage and the decisions behind those tasks harder to understand.
Closing Reflection
The challenge in both lifestyle hospitality and AI adoption is translating judgment into systems without stripping away the judgment that made the experience valuable in the first place.
A property still has to decide who it is for, what kind of experience it wants to create, and which decisions can be standardized without becoming generic. AI does not remove that responsibility. It makes those choices more important because poorly defined processes can now be repeated faster and at greater scale.
Independent hotels have an advantage when they use their size well. They can often make decisions faster, stay closer to their guests, and adapt without waiting for layers of approval. But that flexibility only works when the property is clear about what should remain consistent, what can change, and when a person needs to step in.
The opportunity is to let technology carry more of the repetitive work while giving staff more room for the judgment, attention, and local knowledge that make an independent hotel worth choosing. The best use of AI may be the one guests barely notice, but employees feel every day.