HM on Location: Hotel AI delivers unexpected results

THS 2026 Real-World Use Cases for Implementing AI in Hotel Operations
From left: Moderator Nazpari Aydin, co-founder of Destination AI, Dan Fernandez, Jacqueline DeChamps, Gustaaf Schrils and Keryn McNamara. (Forge)

MIAMI BEACH — Hotel operators are turning to artificial intelligence to reduce labor expenses, increase revenue and improve the guest experience. The good news is the technology is largely delivering results. The interesting news is that those results aren't always what operators expected, according to the participants in Tuesday’s “Real-World Use Cases for Implementing AI in Hotel Operations” panel at the Hospitality Show in Miami Beach, Fla.

Panelist Dan Fernandez, vice president of digital marketing and strategy at Concord Hospitality, points to voice AI as an example. The company deployed this technology at a 200-acre Florida resort where voice AI answered about 42,000 telephone calls over six months. It resolved about 30 percent of those calls without transferring them to an employee. 

That must have moved the needle on labor savings, right? Not exactly.

“It hasn’t necessarily driven crazy labor savings,” Fernandez noted. 

That doesn’t mean the AI initiative was devoid of value. Instead, Concord found that its use freed employees from routine interruptions, thereby improving response times and allowing front-desk associates to devote more attention to the guests standing in front of them.

“If there’s 7,000 questions it answered, that’s 7,000 calls that were diverted from the front desk or the hostess stand, where those individuals almost certainly had a human they were already engaging with,” he continued. “That’s 7,000 times where the phone was not ringing off the hook to the side, which always creates a dilemma, especially if a line builds and they must decide how to address that.”

AI’s Payoff Moves Beyond Labor Savings 

Naturally, Concord isn't the only one finding value in AI's ability to make employees more productive as they combat the seemingly ever-rising expense of labor.

In fact, Aimbridge Hospitality developed its own technology precisely to forecast staffing needs. The company's proprietary Labor Insight & Forecasting Tool (LIFT) compares scheduled labor with anticipated demand, then identifies areas where managers may need to make adjustments.

“We are applying AI to pull forth anomalies and insights,” said panelist Keryn McNamara, Aimbridge's CIO. The goal is to identify the three most pressing labor-related issues requiring managers' attention each day.

Initially focused on housekeeping and laundry, LIFT is now being expanded to incorporate food-and-beverage labor standards.

But even a tool that helps managers make better decisions won't accomplish much if those managers don't use it. That’s why Aimbridge integrated LIFT into the environment its employees already accessed every day. 

“That meant additional security, but less sign-on integration without another environment that they would have to log into, which creates a more seamless experience,” McNamara continued. “We think has led to the over 86 percent adoption rate that we're experiencing with the tool, and because we built it using skills that we had in-house—using our team and our knowledge— there weren’t any additional costs to pass on to owners. It's a huge win in that regard.”

Panelist Jacqueline DeChamps, COO of Towne, described a similar approach using Legion, a workforce management platform that incorporates hotel volume, group business and information about upcoming events. The technology calculates how many employees a property might need, yes, but it does more than that.

“It's not just giving us optimal schedules,” DeChamps said. “It's telling us when we need more part-time employees, more people or a different staffing mix to meet the hotel's service standards.”

Towne also deployed AI during America's 250th anniversary celebrations to identify road closures affecting hotels in Washington, D.C., Boston and Philadelphia. The system flagged upcoming road closures around Towne-served hotels, allowing managers to bring employees in earlier, adjust staffing locations and notify guests that they might need their vehicles before nearby streets closed.

That's a guest satisfaction win as much as an operational one.

Omni Hotels & Resorts is also deploying AI in the hopes of increasing guest satisfaction. Its AI concierge has handled about 1 million guest requests this year, according to panelist Gustaaf Schrils, the company's CIO. Here’s where the system gets interesting: it’s moved beyond answering guest questions and actually initiates the work needed to fulfill their requests.

The system can even process multiple requests simultaneously. “In one text, you can ask for a pillow, a burger and report that the light at the desk is burned,” Schrils said. “It will generate three tickets: one for housekeeping, one for food and beverage and one for engineering.”

The key, Schrils continued, is knowing which tasks should be automated and which still require a personal touch.

“I don't see us checking people in using a robot,” he added. “As long as there is that interaction at the front desk, then use automation to make the rest of their stay flawless and efficient.”

Uncovering Revenue Hiding in Plain Sight with AI

Improving profitability is another area where AI can help. One of the biggest ways it can do this is by identifying ways to generate more revenue from existing assets.

Take parking, for example—“the highest-margin business that you can have within your business,” DeChamps said.

Towne is using AI and machine learning across about 100 hotels to dynamically price parking based on more than 100 data points. Rather than relying on static rates, the technology adjusts pricing to reflect demand, including major events that bring a surge in parking demand.

“Because when Taylor Swift is in town, you absolutely should get more revenue for that parking,” DeChamps said, citing parking margins of about 80 percent. She also pointed to another revenue source: spaces that would otherwise sit empty. Towne is making those spaces available through parking marketplaces, allowing hotels to capture business they weren't getting before.

Fernandez believes the same logic could—and should—be applied to other items a hotel sells. “I think there are massive opportunities for AI to help us manage revenue across the entire experience, as opposed to just rooms,” he said. 

Schrils pointed to an unsold tee time or spa appointment at Omni's resorts. Like guestrooms, these reservation dates are perishable goods. Once the timeslot passes, the chance to sell them does as well.

“Once they're gone, they're gone,” he said.

His suggestion? Use AI to identify those openings, determine which guests already on property might be interested and market the available appointments directly to them, potentially at discounted rates.

Of course, generating more revenue isn't always about selling an additional service. Sometimes it's about not losing a prospective customer in the first place.

With that in mind, Fernandez cautioned that while voice AI performs well with routine questions, it can produce lower conversion rates when handling more complex inquiries from wedding planners, event organizers and other group prospects.

The smarter play, in his opinion, was to deploy AI earlier in the sales process, particularly when hotels receive requests for proposals. He believed AI could help qualify and prioritize those inquiries, send initial responses and identify missing information before a salesperson takes over.

“We know that response time correlates to conversion rate,” he said.

Still, Fernandez emphasized that those early exchanges should lead to a human handoff, giving salespeople the chance to pursue qualified business rather than leaving prospective customers navigating an automated conversation.

Even Good AI Ideas can Fail

Unfortunately, AI is not immune to the notion of “good idea, poor execution.” Even the most promising AI application won't accomplish much if the data behind it isn't reliable. It’s a lesson Aimbridge learned the hard way.

The company rolled out an internal AI assistant designed to answer questions about its hotels. How many properties does it operate in Miami? Who are the general managers? How many rooms do they have?

Despite the questions being fairly straightforward, the technology couldn't reliably answer them.

“It failed,” McNamara acknowledged.

The problem was the underlying data. Aimbridge discovered it needed to restructure that information before the technology could interpret it correctly.

“We thought that it was [AI ready], and it wasn't,” she continued. “It was a fantastic failure.”

Aimbridge has since resolved the problem, but the experience reinforced importance of having accurate, properly organized data before expecting AI to deliver useful results.

Schrils has seen this same type of problem. Asking an AI assistant when a hotel restaurant opens, for example, may produce conflicting answers if the restaurant operates on different schedules during peak and shoulder seasons.

“My recommendation is build trust, and it can so easily be broken,” Schrils said, referring to guests' confidence in the information provided by a hotel's AI systems. “So, keeping your data clean is going to be extremely important to gain that trust.”

This requires operators to pay attention to the accuracy and consistency of the information feeding these systems, especially when guests are relying on those answers.

Getting the technology to work is only half the battle, however. 

“Deploying technology is not the problem,” Schrils said. “The problem is getting it embraced and adopted.”

His recommendation was to involve employees early, giving them a chance to test new applications, provide feedback and have a say in how those tools are implemented.

DeChamps encouraged managers to go a step further with AI. “Ask them to solve something,” she advised.

It’s an approach Aimbridge has already taken. In addition to an AI learning series covering everything from data governance to practical applications, the company established a community of employees experimenting with the technology.

For example, one employee used AI to examine how spending on property improvements affects guest satisfaction scores. The findings were turned into a dashboard that Aimbridge now uses to inform discussions across the organization.

Fernandez is also a believer in hands-on experience. He encouraged employees to move beyond asking chatbots questions to try building personalized AI agents. It’s a good exercise—one that just about anyone can do. “You don't need coding experience or to be a developer,” he added.

Naturally, there's another group that needs to enter the AI chat: the owners, brands and management companies responsible for deciding which technologies to invest in and how they're used.

“I think we're all responsible,” McNamara said. “To solely put it on the shoulders of one entity would be doing the rest a disservice.”

She believed there were two steps any operator evaluating their next AI investment should take. The first is to talk to other organizations already using the technology to determine whether the vendor's claims hold up. The second is to test it before committing to a full rollout.

“I would suggest you pilot,” McNamara continued. “Because you will either have success and great adoption, or you will fail and you will learn a ton.”

Either scenario may bring about some challenges. If you fail, you’re still a good distance away from solving the problem you employed AI to tackle. But what if it does what you hoped it would do? What if it solves the problem. Well, then the challenge is adoption, which first means integration. 

Yes, the first step in recognizing that you’ve solved a problem may be acknowledging that problem in the first place. That’s why many truly successful AI adoptions often require changing how hotels operate, not simply adding another technology to their existing systems.

“Many people want to throw it on the shoulders of IT,” he said. “But the reality is, IT is just a tool. It's the change in the workflows and processes that need to be made to really achieve the benefits of AI.”