A guest can now book, check in, unlock the room, order towels and check out without speaking to anyone. It is fast and it is seamless. Whether it builds loyalty is a different question, and it was the one a panel of hospitality technology leaders took on at ITB 2025.

The session, “From booking to checkout: do digital experiences and AI foster loyalty better than human interaction?”, was moderated by Thorhildur Edda Gunnarsdóttir, CEO at Sweeply, with three panellists:

• Erlendur Steinn Gudnason, Co-Founder and COO of Sweeply

• Frederik Meinhardt Moller, Global Director of Sales at AeroGuest

• Ulrich Pillau, Co-Founder of Apaleo

What did the room actually vote for?

Thorhildur opened by asking the audience to choose: if you had to pick one, does AI-driven convenience beat human interaction at driving loyalty?

Just 12% raised their hands for AI.

That is a striking result from a room full of technology buyers, and it is worth taking seriously rather than explaining away. People still believe personal service is what makes guests return. At the same time, nobody in the room disputed that guests now treat speed, control and personalisation as a baseline rather than a delight.

Erlendur put the reconciliation simply: “AI and automation are all about enhancing the guest experience, not replacing it. You can use it to save time where you’re not adding value and connect with the guest in more meaningful ways.”

Where does AI belong in the guest journey?

The useful question is not whether to automate but which minutes to automate. Stage by stage, the panel’s answer looked like this.

Booking and pre-arrival

Booking used to be transactional. Frederik of AeroGuest argued that it no longer is: “Travellers seek relevance, they want convenience and speed. AI is able to provide that data so people get relevant pre-arrival information and check-in invitations.”

Flexible check-in times, an upsell the guest actually wants, a weather-based suggestion. The bar is relevance, not volume. Sweeply’s AeroGuest integration exists for exactly this reason: preferences collected before arrival are only useful if they reach operations.

Arrival

Check-in is the moment of truth. Frederik was blunt about the version that fails: “You don’t want to be that human robot standing behind the desk, taking in a lot of people, saying ‘next in line’. You want to create real, meaningful relations, and that’s not possible when you also need to fill out registration cards, take care of payments, ask for nationality, date of birth.”

CitizenM came up as the counter-example. Guests check in at kiosks while ambassadors roam the lobby to help, recommend and talk. According to Ulrich, “Guests often say it’s better than a 5-star hotel. The technology does the bad job, and the ambassadors take care of the people.”

Erlendur added the caveat that stops this becoming a universal rule: “If you have a more five-star experience, then obviously there is more personal interaction involved. Think about what you need, choose the right technology.”

In-stay

“You ask for something and it magically appears, and that magic is what hotel operation is really about,” said Erlendur. The magic is routing. A towel request that lands on the right person’s task list, a fault report that reaches maintenance with a photo attached, a stayover that gets scheduled without a supervisor rebuilding the list by hand.

THE FLAG, a six-hotel group in Zurich running on Apaleo, published what this looks like measured: 1,800 reservation traces processed with five misses, over eight hours a week saved, and 80% fewer complaint-related tasks. That is documented in their case study, and the mechanism is described in how AI agents turn PMS reservation notes into housekeeping tasks.

Why does infrastructure decide whether any of this works?

Ulrich was the least diplomatic panellist on this point: “Hotel tech is totally broken. 90 percent of it is legacy systems.”

That matters for loyalty because guests do not experience your architecture, they experience its symptoms: the slow booking page, the check-in that fails at the last step, the preference recorded in one system and invisible in another. Layering a modern guest app over a legacy stack produces friction the guest reads as carelessness.

The work is unglamorous: real integrations, fewer silos, one version of guest data. Sweeply connects to over 20 property management systems, including Apaleo, precisely because operations software that cannot read the reservation is another silo. If you are still choosing a foundation, start with the hotelier’s guide to property management systems.

Where human interaction still wins

This is the part that gets lost when a panel like this is summarised as “AI is the future”.

Service recovery. When something has gone wrong, an automated apology makes it worse. A guest whose room was not ready, whose luggage was lost or whose complaint was mishandled needs a person with authority to fix it. No routing engine substitutes for that, and hotels that automate this stage tend to lose exactly the guests they most needed to keep.

Ambiguity. “The room feels strange” is not a ticket type. Interpreting a vague complaint, or noticing that a guest is unhappy without saying so, is judgement work.

Accessibility and preference. Some guests cannot or will not use an app: no data roaming, low digital confidence, visual impairment, or simply no interest. A digital-only journey excludes them. Keep a staffed, working alternative for every self-service path.

High-touch segments. In luxury, the interaction is the product. Automating it away removes the thing the guest paid for.

Emotional moments. Anniversaries, bereavement travel, a first visit to a city. These are made by a person who noticed.

The 12% vote was not nostalgia. It was an accurate read of where the value sits, and the panel’s consensus was not that AI wins, but that AI should buy back the hours that make the human moments possible.

Checkout is the start of the next stay

Erlendur reframed the end of the stay: “Checkout is also preparing a check-in. It’s not the end of the operation for a hotel, it’s a starting point for a new experience.”

Loyalty follow-up has moved on from points. Relevance is the currency: a message that anticipates the next trip beats a generic offer, and a fast, accurate departure that gets the room back into inventory quickly protects the next guest’s arrival experience. Turning small moments into revenue covers the commercial side of this.

What to do next

Take one stage of your guest journey, ideally arrival, and list every minute a staff member spends on admin that a guest does not value. That list is your automation brief, and it is more useful than any vendor comparison. If you want a second opinion on it, email the list along with your room count and PMS to hello@getsweeply.com and ask which parts Sweeply would remove and which it would leave to your team.

Sources: the ITB 2025 panel “From booking to checkout: do digital experiences and AI foster loyalty better than human interaction?” moderated by Sweeply; THE FLAG Zurich operations and guest complaints case study; and the Apaleo developer documentation for the integration standards discussed.