Ernest, le coéquipier IA des équipes commerciales hôtelières, est désormais disponible pour tous.Demander une démo
Blog

What does it take to make AI work in hospitality?

Two colleagues discuss a spreadsheet on a laptop at a sunlit meeting table.

AI is opening up new possibilities for hotel commercial teams, from simplifying everyday reporting to helping employees make better decisions. But as more hotel groups begin adopting the technology an important question remains. How do you turn that potential into something genuinely useful?

For many hotels, the starting point is automation. It makes sense. Revenue managers spend time preparing forecasts, reviewing performance and gathering information from different systems. AI offers a way to reduce that workload.

But the potential goes further, says Heidi Cosio, who leads Ernest Crews at Lighthouse.

"Automation is a natural starting point when exploring AI. But the real opportunity is to think beyond the tasks we can automate and consider what we want to achieve."

Ernest is Lighthouse's AI teammate for hotel commercial teams, bringing together hospitality-specific knowledge and commercial data in a conversational workspace. Ernest Crews are Lighthouse hospitality specialists who partner directly with hotel groups to make Ernest fit their business. They build bespoke connectors, develop custom skills, and train teams to put AI to work every day.

As Senior Director of Ernest Crews, Cosio works closely with hotel groups navigating these decisions. She also brings years of experience leading hotel revenue teams, giving her perspective from both sides of the implementation process.

The lessons emerging from that work offer a useful starting point for hotel groups deciding how to approach AI. We sat down with Heidi to get her thoughts, and what follows is her perspective on where you should begin when integrating AI at your properties.

Key takeaways

  • A faster forecast isn't necessarily a better one. The real value of AI comes from identifying opportunities and helping teams decide what to do next.

  • Don't roll out AI to everyone at once. Start with a core team, establish shared priorities and develop useful workflows before scaling adoption.

  • New technology won't change old habits on its own. Employees need support to rethink familiar processes and move from routine reporting toward strategic decisions.

  • The right data makes AI more valuable. Connecting commercial, financial and other relevant information can help hotels make more holistic decisions rather than focusing on room rates alone.

Start with the decision, not the automation

Forecasting is one of the first tasks hotel groups want to automate. But before deciding how AI should prepare a forecast, it's worth considering what happens once it's finished.

"One of the most common requests we get is for AI to prepare forecasts," Cosio explains. "But what I want to understand is what happens once that forecast is updated. Did you identify a gap? And when you identified that gap, what was the next step you took? That's where we want to help people connect the dots, from producing the forecast to deciding what to do with it."

A forecast might reveal weaker-than-expected demand for a particular period. The team then needs to investigate the shortfall and decide how to respond, perhaps through marketing activity, group business or changes to its pricing strategy.

The same applies to marketing. Rather than concentrating on automating campaign reports, a hotel might want to understand where to invest its budget, when to launch a campaign and how to assess the results.

Working backwards from these decisions helps clarify what information AI needs and which activities are worth automating.

“Start with your endgame, rather than starting with your automation.”

Heidi Cosio, Senior Director of Ernest Crews at Lighthouse

You can also reveal opportunities to improve an entire process, particularly where reaching a decision involves several departments or sources of information.

Group displacement analysis is one example. Assessing a potential booking may involve reviewing tentative business, historical ancillary revenue, the costs associated with accommodating the group and its impact on the hotel's forecast.

The team might then consider alternative dates, negotiate different terms or prepare a response to the client.

Ernest Crews have been exploring how to bring more of these activities together, drawing on additional information from hotel systems to support a more comprehensive assessment.

For hotels developing an AI strategy, automating existing processes can deliver immediate benefits. But starting with the outcome they want to achieve can reveal opportunities that might otherwise go unnoticed.

"Start with your endgame, rather than starting with your automation," Cosio advises. "What are you going to do with it once it's done?"

Revenue and group-sales managers assess a hotel meeting room while discussing a potential group booking.

Establish a shared direction before scaling adoption

Giving employees access to AI and encouraging experimentation can be a useful way to discover new applications. For hotel groups, however, there comes a point when individual experimentation needs to fit into a wider approach.

One hotel group working with Ernest Crews recognized this early.

Rather than giving approximately 200 employees immediate access, its leadership established a smaller team to identify priority applications and develop shared routines before introducing AI more widely.

These included processes for distributing consistent information across the organization, giving employees a common starting point while allowing them to explore the results further.

The approach also helped establish who would make decisions about which applications to develop and how they should be used.

"You need an executive sponsor who can help set the direction and make decisions," Cosio explains. "When you have 10 or 15 people involved in a pilot, you're naturally going to have different opinions and priorities. Without someone who can bring those together, it can become difficult to move forward."

An executive sponsor provides a point of accountability, particularly when competing priorities need to be resolved or changes affect several departments.

But the strategy should also reflect the experience of employees who understand how the business operates.

Revenue management, sales and marketing all bring different perspectives, with operations and technology teams involved where relevant. A process that seems straightforward from one department's perspective may depend on information or decisions from another.

Bringing these people together early can help identify those dependencies before new workflows are developed.

Find out what employees actually need

Once the right people are involved, the next step is to understand how they work.

Ernest Crews approach this through workshops in which employees map their daily, weekly and monthly activities, including the reports they produce and the people who rely on them.

"We sit down and talk through their daily workflows. What are you doing every day, every week, every month? Why do you do that? Who gets that information, and why is it valuable to them? We put everything on a shared board, and then we ask the team to identify and rank their five most important priorities."

The exercise can reveal that a report takes hours to prepare but contains information already available elsewhere. Another might serve several departments and be an obvious candidate for improvement.

It can also expose tasks that continue largely because they've become part of the routine.

Rather than attempting to automate every activity identified, the team concentrates on a small number of priorities.

There's also value in helping employees understand what AI can already do before investing in more specialized development.

For everyday questions and analysis, Ernest can be used conversationally, without creating a dedicated skill for every task. More structured requirements, such as producing a weekly report in a consistent format, may benefit from an existing template or a customized skill.

"We do a lot of side-by-sides and training," Cosio explains. "We had someone who needed to produce an important report every week. So we worked through how they could give Ernest the format they wanted and reference that each time. And for something more complicated, we can build a skill with more detailed instructions."

By working through these practical applications with employees, teams can build confidence in AI, establish more effective workflows and focus development effort where it will deliver the greatest value.

Hotel commercial team prioritizes workflow cards on a shared board during a planning workshop.

Prepare people for a different way of working

Even when the benefits of AI are clear, changing established working practices can be difficult.

Hotel commercial teams have spent years developing processes for compiling data, reviewing reports and making recommendations. Employees know where to find information and how to use it.

That familiarity has value. It also means a new approach can initially feel less comfortable, even when it offers significant advantages.

AI can make this transition particularly challenging because employees may associate their expertise with activities that are becoming easier to automate.

Revenue managers, for example, often take pride in their ability to compile detailed reports and interpret the results. If AI takes on more of that work, they need to understand how their responsibilities will evolve.

“I think one of the biggest gaps is that we're not taking away the old and only providing the new to work with.”

Heidi Cosio, Senior Director of Ernest Crews at Lighthouse

Preparing employees for these changes shouldn't begin on the day the technology is introduced.

Before joining Lighthouse, Cosio helped a large hotel management company introduce business intelligence technology across approximately 1,000 properties in under 60 days.

The deployment was fast, but the preparation began three months earlier.

"We started having conversations about change management three months before deployment. We talked about what the change would look like, how people would feel about it and what success would look like for them. They already had a framework they knew really well. They knew where to look and how to find what they needed, so we had to prepare them for a completely different way of working."

They also eventually retired the old reporting workflows, ensuring employees weren't simply returning to familiar processes.

"I think one of the biggest gaps is that we're not taking away the old and only providing the new to work with."

This doesn't mean hotels should immediately abandon systems that are working. The new approach needs to be reliable, employees need appropriate training, and critical processes must continue to function.

But it does mean leaders need to think seriously about how AI becomes part of normal operations.

Of course, replacing familiar workflows isn't always straightforward.

Reporting is a good example. Hotel teams often prepare information in a particular format because that's how it has always been presented to senior leadership or ownership. Those expectations are important, but they can also discourage employees from exploring whether the information could be delivered more effectively.

"It's worth understanding what we're ultimately trying to achieve with a report and whether there might be a better way to deliver that information," Cosio explains. "If we can make it easier for an owner to find what they need, or provide additional insight without requiring them to work through all the details, that's an opportunity to add real value."

The distinction is between preserving the information stakeholders need and preserving every step involved in delivering it.

Some requirements, particularly those involving financial accuracy or contractual obligations, aren't negotiable. But the way information is organized, analyzed and presented may offer opportunities for improvement.

Making those changes can require conversations that employees aren't always in a position to initiate themselves. That's another reason executive sponsorship is so important: leaders can help teams challenge established practices, agree on what needs to change and ensure the new approach still meets the business's requirements.

Focus on the information that matters

Changing established workflows is only part of the adjustment. Commercial teams also need to become comfortable with a different way of reviewing information and making decisions.

A revenue manager might be accustomed to examining a report covering an entire year of performance. AI could instead highlight the five dates that require attention, allowing the manager to concentrate on understanding those opportunities and deciding what action to take.

But when employees are used to checking every figure themselves, a shorter summary can initially feel insufficient.

"You're changing that automated workflow into a conversational sparring partner," Cosio explains. "Instead of going where your eyes are used to going, you have to get comfortable with the idea that you don't need to see 100% of the data. You might only need to look at five days to make a strategic decision."

Teams still need to understand where AI-generated information comes from, investigate discrepancies and verify the figures that matter to a decision.

The adjustment is learning to separate information that genuinely requires closer examination from detail that employees are accustomed to reviewing out of habit.

The objective isn't to replace commercial judgment. It's to make that judgment more effective by directing employees' attention toward the issues where their experience can make the greatest difference.

A hotel commercial team member reviews data tables and charts across two monitors in an office.

Better decisions require better commercial context

As hotel groups explore more sophisticated uses of AI, the information that’s available becomes increasingly important.

General-purpose AI tools can already help with drafting communications, summarizing documents, analyzing uploaded information and other everyday tasks. With appropriate integrations, they can also support more specialized analysis.

But they don't automatically have access to a hotel's commercial information or the context needed to interpret it.

That context becomes important when AI moves beyond answering questions to helping teams evaluate commercial decisions. Before reducing room rates, for example, a hotel could use AI to identify ten alternative actions worth considering.

"How far can you go before simply dropping your rate?" Cosio asks. "You've got to do bigger things, bolder things. You've got to pull in more aspects of the business and make holistic decisions rather than siloed decisions."

Evaluating those alternatives requires an understanding of the hotel's performance, market conditions, competitive position and commercial objectives.

Hospitality-specific AI can provide a useful foundation by combining relevant expertise with access to commercial data.

Ernest, for example, draws on relevant Lighthouse data, including pricing, demand, performance and competitive information, to help employees investigate opportunities at their properties and verify the information behind recommendations.

“How far can you go before simply dropping your rate? You've got to do bigger things, bolder things.”

Heidi Cosio, Senior Director of Ernest Crews at Lighthouse

But hotel groups increasingly want to incorporate information that exists over and above traditional commercial systems.

Financial data, general ledger accounts, profitability analysis and guest feedback are among the sources being considered. These could help teams understand not just how their rooms are performing, but how different factors influence the wider business.

"The amount of data hotel groups want to bring in is fascinating," Cosio says. "They're looking at financial information, general ledger accounts, profitability analysis and guest review scores. They want to understand how their rankings are changing based on guest sentiment, how that relates to their pricing and what they might do differently as a result."

The extent of that fragmentation has been greater than expected.

"I don't think we appropriately measured how fragmented that data was, or the scale of it," Cosio acknowledges.

Even with core commercial data readily available, hotel groups may rely on numerous additional systems containing information they want to incorporate into their analysis.

That creates both an opportunity and a practical challenge for hospitality-specific AI. Bringing together more relevant information could support more comprehensive commercial decisions, but the ability to connect existing systems becomes an important consideration when choosing a solution.

It also raises a broader question about the complexity of hotel technology. As groups explore ways to bring their information together, some may begin reconsidering how many separate systems they need in the first place.

Security and privacy deserve the same attention.

Hotels should establish where their commercial information will be stored, who can access it and whether it can be used to train shared AI models.

With Ernest, customer data is kept separate from other customers' information and isn't used to train shared models.

Whatever solution a group chooses, employees need to be able to verify the information supporting its recommendations. A persuasive answer is of limited value if its underlying data is incomplete, outdated or unsuitable for the decision being made.

Measure success by what changes in the business

Time savings are among the easiest benefits of AI to identify.

A report that once took an hour might be prepared in minutes. Employees could review performance across several properties without manually gathering information from different systems.

These are worthwhile improvements, and they're relatively straightforward to measure.

But hotel groups should also assess whether their original objectives are being met.

In Ernest Crews' early engagements, success criteria have included data quality, reporting timeliness and whether agreed connections to other systems are working.

These measures help establish whether the technology is reliable enough to support everyday work.

Commercial outcomes require a different assessment.

A hotel group using AI to improve marketing decisions might examine changes in campaign targeting, expenditure and returns. Another looking to improve portfolio oversight could evaluate how quickly its teams identify and respond to emerging risks.

Attributing financial improvements to AI won't always be straightforward. Market conditions, employee decisions and other initiatives can all affect the results.

That makes it important to agree on relevant measures before implementation, rather than trying to demonstrate value retrospectively.

There is also a broader consideration: how AI changes what commercial teams are capable of doing.

"If I were a customer myself, I'd want to understand whether my team is developing strategically," Cosio says. "Are they getting faster, more efficient and smarter?"

A team that spends less time preparing routine reports may be able to support more properties, investigate opportunities in greater depth or work more closely with colleagues in other departments.

These changes won't necessarily appear immediately in a financial report. But they can influence how effectively the business uses its existing expertise and resources.

For hotel leaders, this raises questions about the skills their teams will need and how responsibilities might evolve.

Employees who have traditionally concentrated on reporting and analysis may have opportunities to contribute more to commercial planning, while managers may need to place greater emphasis on developing their strategic judgment.

The aim isn't to turn every revenue manager into an expert in every department. It's to give employees the capacity and understanding to consider the wider implications of their decisions.

Two colleagues review a laptop spreadsheet and handwritten notes during a discussion

Making AI work for your hotel group

For hotel groups considering their next steps with AI, three questions provide a useful starting point.

Where are the gaps in our existing workflows? Look at how your team works today, what information it needs and where time or opportunities are being lost.

Is our team ready to work differently? Consider whether employees have the skills, support and willingness to adopt new processes. Think about how their responsibilities might change as routine work becomes easier.

What are we ultimately trying to achieve? Define the business outcome, understand which systems and resources are needed, and make sure your budget reflects the ambition of the project.

Those answers will influence the technology they need, the people who should be involved and the level of investment required.

General-purpose AI may be sufficient for some everyday productivity tasks. More specialized applications involving proprietary commercial data, additional connections and organization-wide workflows may require greater resources.

There can be a place for both, depending on what the business is trying to accomplish.

It's also worth considering how those requirements could evolve. A solution that addresses an immediate need may be valuable today, but the group's longer-term objectives could require a different approach.

Producing reports faster is one thing. Giving commercial teams a clearer view of their business, helping them identify opportunities and making it easier to act on those insights can deliver far greater value.

"We need to think bigger about what we're trying to accomplish," says Cosio. "It's not just about adjusting rates or improving one part of the business. We need to bring more things together, look at the bigger picture and make decisions that take the whole business into account. That's where I think the real opportunity is."

Frequently asked questions

What is Ernest, and how can hotel commercial teams use him?

Ernest is Lighthouse's AI teammate for hotel commercial teams. He brings together hospitality-specific knowledge and commercial data in a conversational workspace, helping employees investigate performance, identify opportunities and make better-informed decisions. Teams can ask questions in natural language and explore recommendations grounded in relevant Lighthouse data.

What are Ernest Crews?

Ernest Crews are specialist Lighthouse teams that work with hotel groups to tailor Ernest to their business. They help identify priorities, develop customized skills and routines, connect relevant systems and support employees as they adopt new ways of working.

How is Ernest different from general-purpose AI tools?

General-purpose AI tools can help with tasks such as drafting communications, summarizing information and analyzing documents. Ernest is built specifically for hospitality, with access to Lighthouse's commercial intelligence and an understanding of hotel revenue management. This allows him to provide recommendations informed by property performance, demand, pricing and competitive conditions, subject to the data available to each customer.

Can Ernest work with a hotel group's existing systems?

Yes. Ernest draws on Lighthouse data and can work with supported third-party integrations. Through Ernest Crews, hotel groups can identify additional information they want to incorporate, such as financial or operational data, and explore the connections needed. The availability of specific integrations depends on the systems involved and the implementation requirements.

Can hotel teams verify Ernest's recommendations?

Yes. Ernest is designed to make his recommendations transparent by linking them to the underlying Lighthouse data and supporting information. This allows employees to investigate the figures behind an insight, assess its relevance and apply their own commercial judgment before deciding what action to take.

How can Ernest Crews help hotel groups adopt AI successfully?

Ernest Crews work with hotel groups to understand their existing workflows, identify valuable applications and establish shared priorities. The team also supports training and change management, helping employees move beyond familiar reporting processes and use AI to support more strategic commercial decisions.

Author image

Joe Hanly

Author image

Joe Hanly

Joe Hanly is a writer and content creator at Lighthouse, helping hoteliers navigate industry trends with clear, engaging storytelling. He believes the best insights come from real experiences – that’s why when he’s not writing about travel and hospitality, he’s out exploring it firsthand. It’s all research. Sort of.

Put AI to work across your hotel group. Build tailored workflows and scale AI with Ernest Crews.