Putting AI to work in a 100-room hotel without letting anyone go
No staff were retrenched. Over three months, five-star Google reviews rose 22% on the same period before, advance sales 17% and restaurant sales 41% with room service restored, and staff satisfaction rose by more than 50%. All measured against the hotel's own baseline.
Anonymous client case study. Results reported by the owner against the business’s own baseline.
Illustrative scenes in the video are labelled “Recreated demo”.
Putting AI to work in a 100-room hotel, without letting anyone go
At a glance
- The hotel: three-star, 100 rooms, with a restaurant.
- The mandate: take the repetitive, round-the-clock work off the team without cutting the team.
- What we built: three AI voice agents (front desk, reservations, room service), automated package quotes, a no-download guest interface, and well-timed review requests.
- The results, against the hotel's baseline over three months:
Measure
Change
Staff retrenched
None
Five-star Google reviews (on the same period before)
+22%
Advance sales
+17%
Restaurant sales
+41%
Staff satisfaction
More than 50%
The problem was never effort
The team at this hotel was working hard. The trouble was where the effort went.
The front desk answered the same questions all day and all night: where the hotel is, how to get there, what a room costs, whether one is free. Each call pulled someone away from the guest standing in front of them.
Reservations served tour agents around the world, independent travellers, small private groups and corporates, each working in their own time zone. An enquiry that arrived at 2am was answered at 10am, and by then the buyer had often booked somewhere else.
The restaurant had given up on room service altogether. It was short-handed, and some of the team weren't confident taking orders in English, so management judged the risk of getting orders wrong to be too high. Every night, a hundred rooms of potential diners went unserved.
One section at a time
We didn't switch everything on at once. Each agent went live, settled, and earned the team's trust before the next one began.
1. The front desk agent.
- What it handles: every call, at any hour, about directions, location, rates, availability and the standard questions any caller asks a hotel.
- What it doesn't do: replace the front desk. The routine phone traffic disappeared from the desk, and the people stayed. Their time went back to guests in the lobby and to the problems that genuinely need a person.
2. The reservations agent and automated quoting.
- The groundwork: a knowledge base of room types, rates, rules and packages.
- The process we automated:
1. The enquiry is taken.
2. The right package is put together.
3. It is formatted.
4. It is sent, immediately, with no waiting for someone to reach the office.
- The result: confirmations could come back while the buyer was still interested.
- Where people come in: custom packages stay with the reservations team. The AI suggests options based on the guest's profile, and a person puts the package together and takes it from there.
3. The room service agent.
- What it knows: the full menu and how ordering works.
- How it works: guests order by talking to the agent, in English or Malay, which takes the order clearly, so language no longer decides whether the hotel can sell a meal.
- The result: room service came back. Because guests were now in regular conversation with the hotel throughout their stay, restaurant sales didn't just recover. They rose 41%.
No app to download
Guests reach all of this without installing anything:
- How they get in: they scan a QR code or tap an NFC tag in the room or around the property.
- What they can do: an interactive screen lets them talk to the front desk, order room service or make a booking.
- Why it works: with no download and no sign-up, guests actually use it.
Asking for reviews at the right moment
So many guests were now talking to the hotel that we could automate the conversations themselves, and resolve issues within them.
That changed how we asked for reviews:
- What most hotels do: send a fixed "please rate us five stars" message on a schedule.
- What we did: show the review request to guests at the moment they're happy: while they're still staying, or just after they leave.
- When a guest complains: we don't send the request. The issue is escalated to the duty manager, who calls the guest and listens. Guests leave bad reviews when they feel nobody listened. They remain free to review the hotel on Google whenever they like.
The hotel received many more reviews. Five-star Google reviews rose 22% compared with the same length of time before.
What the team does now
Team
Before
After
Front desk
Answering repetitive calls
Present with guests, solving problems
Reservations
Quoting every enquiry by hand
Handling the complex, high-value bookings, each from a complete brief
Restaurant
No room service
Room service restored, with the language barrier removed from ordering
No one was retrenched, and staff satisfaction rose by more than 50%, measured with simple polls and feedback forms before and after. The work had been tedious and repetitive; now it is more fulfilling.
Before the project, the team's worry was that AI would replace them. Afterwards, they said they could finally focus on guests: listening to them and really talking to them. This is a three-star hotel, and guests now feel it genuinely cares about them. That is worth more than any star rating.
What we'd tell another hotel owner
Start with the calls. Front-desk phone traffic is the easiest win and the fastest relief for your team.
Build the knowledge base first. An agent is only as good as the rates, rules and menus behind it.
Go one section at a time. Let each part settle before you add the next.
Ask for reviews when guests are happy, and when they're not, listen first. Escalate complaints to a manager who calls back.
Decide early that no one loses their job. Then tell the team. It changes how they receive the change.
The engagement is ongoing. Further projects are under way at the same hotel.
Read the transcript
We put three AI agents into a hotel. Nobody lost their job. It's a three-star hotel, and its guests now say it really cares about them. Here's exactly how we did it, step by step. We didn't start with AI. We started by listening to the phones. The front desk took more than a hundred calls a day, and about ninety percent were the same handful of questions: where are you, how do I get there, what's the rate, do you have a room. That told us where to begin. An AI agent is only as good as what it knows. So before switching anything on, we built the hotel's knowledge base: room types, rates, booking rules, packages and the full menu. Most "AI doesn't work" stories fail right here. Each agent took about fourteen days to build, demo and test before going live, one at a time. First: the front desk. It answers every call, any hour. The front desk team didn't disappear. Their time went back to the guests standing in front of them. Reservations had a 2am problem. Tour agents and travellers abroad sent enquiries overnight, got answers the next morning, and booked somewhere else. So the second agent takes the enquiry, builds the right package and sends the quote straight away. Here's the expert part: deciding what the AI handles, and what goes to a person. Custom packages stay with the reservations team. The AI suggests options based on the guest's profile, and a person puts the package together and takes it from there. The restaurant had stopped room service: short-handed, and the team weren't confident taking orders in English. So the third agent knows the whole menu and takes the order clearly, in English or Malay. Room service came back, and restaurant sales rose forty-one percent. Guests reach all of it by scanning a QR code or tapping their phone. No download, no sign-up, so they actually use it. Here's what we learned. Guests leave bad reviews when they feel nobody listened. So the review request only goes to guests who are happy, at the right moment: while they're still with us, or just after they leave. If a guest complains, we don't send it. We escalate it, and the duty manager calls them and listens. They can still review us on Google any time; we just don't push it while they're upset. Five-star Google reviews rose twenty-two percent on the same period before. The team's first worry was being replaced. Instead, the boring, repetitive work moved to AI, and they got to do the part that matters: listening to guests. Staff satisfaction rose by more than half. Nobody was retrenched. In three months: five-star Google reviews up twenty-two percent. Advance sales for future dates up seventeen percent. Restaurant sales up forty-one percent. Staff satisfaction up more than half. And zero retrenchment. Buying AI is easy. Knowing what to automate, what stays human, and in what order: that's the job. If you run a hotel, restaurant or service business, message me on LinkedIn, and I'll show you how it would work in yours.
