Hotels Bet on AI Back Offices, but Only 5.6% Budget the Savings
Only 5.6% of hotel companies book AI benefits into budgets or P&Ls, even as 37% plan to raise AI spending by more than 10% next year and operators pitch AI-native back offices.

Just 5.6% of hotel companies commit AI benefits to a budget or P&L (profit-and-loss statement). Yet 37% expect to increase AI investment by more than 10% next year, according to the State of Hotel AI survey. That gap now shapes hotel back offices.
U.S. RevPAR (revenue per available room) is forecast to grow 4.4% this year, per CoStar and Tourism Economics. Spend time with operators at this fall's industry conferences and the mood reads differently: margin defense, rising expenses, tech creep, and owners scrutinizing every line below revenue.
At the Destination AI summit in Washington, D.C., some operating models went well past AI helping accountants work faster. They envision AI-native operators running the back office with agents, then taking responsibility for the financial results. Owners, operators and auditors still need to see how those numbers were produced.
That is the question every operating model must now answer: when AI agents do most of the back-office work, who owns the numbers?
Why owners listen to AI-native operators
For years, owners bought technology to eliminate manual work and ended up with stacks of point solutions, each carrying its own subscription and its own version of the numbers. The new pitch is tidier: one party does the work and answers for the outcome.
Adoption is not the obstacle. According to the State of Hotel AI survey, 44% of hotel leaders use AI more than 10 hours a week, and 90% say it has improved time spent on routine tasks.
Who's checking the agent's work?
Connecting systems and trusting what comes out of them are different things, particularly when hotel data sits scattered across tools that do not speak the same language.
Trust starts with consolidation:
- one system of record with a universal chart of accounts
- a global vendor list
- consistent definitions for key metrics
The industry is not there yet. Roughly 48% of hotel leaders do not believe their data is ready for their AI goals.
Even with unified data, someone still owns the accuracy, backed by a complete record of every action, human or AI. Owners are rightly asking who, by name, is accountable for each agent's work, and whether they can see a full record of what it did.
Where the AI belongs in the back office
AI should carry the transactional load:
- coding invoices and automating payables
- reconciling bank and PMS (property management system, the platform holding bookings and guest folios) activity
- flagging unusual activity before month-end
It should also answer financial questions in plain English, and let teams build their own agents on the AI models they choose, which frees controllers and accountants to investigate exceptions, explain the numbers, and advise the operators and owners who rely on them.
The placement rule is non-negotiable. AI works inside the system of record, drawing on the same data the controller, CFO and owner see, under the same permissions and controls. Every action, human or AI, stays traceable to an accountable person.
Does the system survive a change of operator?
A traditional management company, an AI-native operator and an owner's asset managers all need one source of truth everyone can see, and one that stays with the hotel if the operating model changes.
HIA, the firm putting the argument, describes its own approach as "not taking accountants out of the equation, but elevating them to the work that requires judgment."
Can the savings be booked?
That is the harder test. If AI savings cannot be traced, they cannot be proven, and unproven savings do not survive a margin conversation. Only 5.6% of hotel companies commit AI benefits to their budgets or P&Ls.
The argument is no longer whether AI belongs in the back office. It is whether AI investment gives owners and operators more visibility into their numbers, or adds another layer of technology between them. The best ones reduce the systems between people and their numbers.
The next budget cycle will settle it: either AI savings show up on owner statements, or they remain line items nobody can trace.
More from Rebecca Stone
Show full bio
Senior reporter covering technology and product news in Restaurant and hotel operations at The Pass Brief.
711 articles


