The Year of Operational AI: What Marriott and Hilton Figured Out First

Skift founder Rafat Ali opened the company's Data + AI Summit in New York by calling 2026 "the year of operational AI." For the early adopters, AI is out of the sandbox and inside the daily work of running a hotel. Consumer-facing features get copied fast, Ali noted. The real work is happening in the orchestration layer, the layer that connects AI models to the systems that actually run hotels.

Most of the industry is not there yet. According to McKinsey research, fewer than one in five large travel organizations had fully scaled a generative AI project. Seth Borko, Skift's head of research, has a name for the distance between AI announcements and AI results: "the say-do gap."

The most useful part of the event was hearing the hotel leaders who run this technology at Marriott and Hilton explain how they crossed that gap. The wins they shared rest on a foundation any hotel leader can learn from, including, maybe especially, the management companies operating hotels for owners.

Key points

• Skift framed 2026 as the year of operational AI, with the technology moving out of pilots and into daily operations.
• Marriott's and Hilton's data leaders agreed that AI delivers its full value when two things are in place: a connected data foundation underneath it and real change management around it.
• Marriott's lesson: do not just automate the old process, re-engineer it. The biggest gains come from sitting with the people who do the work and seeing what could be done better.
• The same playbook can work across the industry. For hotel management companies, that means connected labor, operations, and finance data across every property, and a team ready to use what gets built.

Hilton: build the foundation

Michael Leidinger, Hilton's CIO, has spent eight years rebuilding the company's technology from the ground up: breaking apart old monolithic systems, building a modern API and data layer, and moving as much computing as possible to the cloud. None of it was framed as AI at the time. It was about being ready for whatever came next.

The payoff shows up in the numbers. A reservation system that peaked at around 120 million transactions a day now processes over 1.5 billion on busy days. And what matters for operators is what rides on that volume: rate recommendations, revenue management reads, work the old system could never carry.

Data, as Leidinger put it, is "the lubricant of AI." With the foundation set, the visible wins followed. Hilton's AI trip planner, built on Anthropic's Claude and trained on the company's own property content, lets a guest describe the trip they are imagining instead of filling in dates and room counts. Live since March, it is lifting conversion and joining the Hilton Honors app this month.

When Hilton buys rather than builds technology, Leidinger said, the approach is to find a partner with a strong, scalable product and customize it to deliver the full Hilton experience. A scalable product meets the needs of every hotel. The customization delivers what only Hilton needs.

Closer to the front desk, Hilton is testing a conversational agent that could check in a group or coach a new hire through a job with constant turnover. Leidinger's bar for all of it: deliver real value to the owner and franchisee, or it does not matter.

He is just as disciplined on cost. Everyone is talking "tokenomics" right now, he said, and part of his job is teaching people when not to reach for an expensive model. He counts that as change management, too.

Marriott: start re-engineering

If Hilton's story at the event was about the foundation, Marriott's was what you can build on top of it.

Marriott has been running AI pilots for years now, trying things out and learning as it goes. Today, Colin Coleman, Marriott's SVP of enterprise data, analytics, and AI, describes the current phase as "putting points on the board." Coleman joined a year ago after a career in data-heavy businesses, including Equifax, and found much of what he needed already in place in the organization. The work was connecting the dots.

Their AI rollout includes three layers: Copilot for nearly everyone working in their corporate offices, no-code tools at the team level, and a handful of industrial-strength systems aimed at revenue and cost. Marriott started where the risk was lowest, with internal cost savings initiatives, and is now stepping into higher-stakes work that faces the guest.

Marriott's advantage starts underneath the AI. They had enough gains on the table with technology before generative AI entered the room, Coleman said, just from connecting the dots of the customer and the operation. The trick now is connecting that data in a way modular enough to provide optionality, so the same foundation supports whatever you build next. Gen AI poured gasoline on the fire.

Coleman's biggest idea runs counter to how most of the industry thinks about AI. Many assume the role of AI is automation. He sees something bigger. Sit next to a hotel manager for a day, he said, and observe all the spreadsheets and manual steps that have been historically required. They work that way because they have to, not because they chose to. Automate the workaround, and you just get a faster workaround, he said. The better move is to step back, ask what people are really trying to achieve, and redesign the work with them.

The consistent theme: change management

AI technology is advancing faster than most organizations can absorb it. Session after session, from hotels to airlines, the conversation returned to the same conclusion: how much of AI's value you capture depends on change management.

Coleman lives this every day. His job is harder than a pure technology role, he said, because he has to live inside the business. Try convincing a property operator who has manually forecast labor for 20 years that a new model can beat his approach. Change takes more than accuracy. It takes understanding the "physics," the incentives, and the politics of how a hotel actually runs. The build, Coleman said, is just one step. Upskilling the workforce is the part that brings innovation to life.

Leidinger agreed: "The change management is crucial to helping drive this."

The research agrees. Eight in ten travel employees say their company is not ready for AI, and Borko's read is that AI readiness looks less like a software rollout and more like an organizational restructure. The employees most optimistic about AI are the ones who use it daily. The most worried are the ones who never touch it. So put the tools in people's hands early and let the optimism build on its own.

What this means for hotel management companies

Everything Marriott and Hilton described rests on two ingredients: connected data and people brought along with the change. The same two ingredients decide outcomes across a management company's portfolio.

And a management company has its own version of the work, the part that wins and keeps management contracts with owners. Labor planning across dozens of properties. Owner reporting that pulls from every system in the portfolio. P&L visibility across brands and markets. Apply the same playbook there, connect the data underneath the work and re-engineer it with the people who do it, and the gains compound across every property you operate.

The starting point is the same diagnosis that the leaders on stage gave. The word that kept coming up was "fragmented": data scattered across systems that do not talk to each other, with no single view of the operation. A portfolio that spans many properties, brands, and markets knows that challenge well. Coleman said it directly in his closing advice. It all comes down to understanding your customer, and that means having all of your data connected. "I don't care how small or big you are," he said. "That's just the basics of it."

The foundation Hilton spent eight years building, connected data and people ready to use it, is the same prerequisite at portfolio scale. The good news for management companies: it is the part you do not have to build from scratch. Connected hotel operations platforms exist for exactly this. When your labor, operations, and finance data already live in one place across every property you operate, you do not need an enterprise cloud team to reach the starting line. What you still need is a plan to bring your people along.

The pattern across the event was consistent. The companies pulling ahead built the foundation first, then brought their people along. The features everyone sees are what that work makes possible. Rafat Ali called this the year of operational AI. The playbook Marriott and Hilton shared works at portfolio scale, and the gap between the companies doing this and the companies talking about it compounds from here.

Three things to take back to your team

• Audit your data first. If your systems do not talk to each other, no AI capability will save you.
• Treat adoption as the project, not the afterthought. Put the tools in people's hands early.
• Ask what to re-engineer, not just what to automate. Automating a workaround only makes the workaround faster.

Where to start

You do not need a multi-year infrastructure budget to put this playbook to work. You do need to know how fragmented your stack really is, then start closing the gaps that block everything else. Actabl brings your labor, operations, and finance data into one connected view across every property you operate. We are here to help you, as your partner, reach your goals with AI. Request a call with our team to learn more.