Your Staff Already Adopted AI. Your Hotel Didn't.

BCG’s fourth annual AI at Work survey — close to 12,000 frontline employees, managers, and leaders across more than a dozen markets — landed in June with a number that should end a two-year-old argument in hospitality.

74% of frontline employees are now regular AI users. Up 23 percentage points in a single year. BCG has a name for the barrier that just broke: the silicon ceiling, the long-held assumption that AI was a tool for the head office and the knowledge worker, and that the people actually doing the work would never pick it up. That ceiling is gone. Adoption, in the sense most hoteliers have been anxiously measuring it, is over. Your people are using it.

And here is the number that lands directly on the desk of every GM reading this: of those frontline regular users, 42% report saving eight hours a week — a full workday. Meanwhile, 66% receive limited or no guidance on what to do with the time they save, and more than half say they are not reinvesting it into more strategic work.

Read those two sentences together. Your staff are finding a day a week. Your hotel is not capturing it. It is leaking straight out of the building.

Hospitality is not behind. It’s exactly average.

The industry’s own numbers say the same thing from the inside. Canary Technologies’ Catherine Donaldson, writing in the 2026 Hotel Yearbook, reports that 71% of hoteliers say AI is already significantly impacting operations — up ten points year over year — more than 80% plan to increase AI adoption, and only 1% expect to reduce investment.

Near-universal intent. Near-zero retreat. So why does it still feel like nothing has fundamentally changed at the property level?

BCG’s other June/July publication — the agentic scaling playbook from Mark Abraham and Neveen Awad — answers this with the most uncomfortable statistic of the lot. Only 5% of companies are truly agent-first. And, in Abraham’s words, “it’s not a spectrum, it’s two clusters.” The leaders have driven adoption above 80% of basic AI tools across the organization. The other 95% sit at “around 30% or less for even the basic tools,” patchy at that, because they “haven’t given employees the latitude to even experiment.”

Two clusters. Not a curve. There is no comfortable middle where you are gradually getting there.

Awad supplies the reason, and it is the single most useful sentence a hotelier will read this year:

“70% of getting to scale is people and change, 20% is data and technology, and only 10% is the algorithms.”

Which means: essentially everything hospitality has been arguing about — which PMS, which vendor, which model, whose agent — is the 10%. The AIHA’s own member survey found precisely this frustration, members “hungry for practical guidance” and pointing at the gap between AI promises and operational value, compounded by fragmented systems and uneven readiness. They’re not asking for better technology. They’re asking for someone to tell them how to change how the hotel works.

Why your pilot died

Every operator reading this has a dead pilot. Here’s the autopsy, and it’s rarely the software.

Are Morch’s piece on Horst Schulze names the pattern exactly: hotels abandon AI investments prematurely because they measure the wrong outcomes on compressed timelines. His expectation-setting is worth writing on a wall — operational relief in roughly 90 days, deeper returns in three to six months, compounding loyalty gains over the long term. Judge a pilot at week six on a cost-cutting metric and you will kill something that was working.

BCG’s data explains the mechanism underneath that impatience. Only a third of frontline employees say leadership’s communications about AI are clear, and only 28% see a strong connection between what leaders say and what the organization actually does. A pilot launched into that vacuum isn’t a pilot. It’s a rumour with a login.

Add the training gap: 72% say the skills expected of them have shifted; only 36% feel they’ve been adequately upskilled. And BCG’s finding on who actually scales — “the trailblazer organizations are also the ones who had trained more than half their workforce.” Awad again: “Transformation happens where the work happens.”

Then the two killers Abraham and Awad name directly, both of which will be familiar to anyone who has watched a hotel group’s AI initiative stall:

• Governance chaos. “Unclear ownership creates a lot of ‘I own that, no I own that’ confusion that stops everything.” Half of BCG’s respondents say their company lacks clear governance for managing teams of people and AI together.
• Not moving from pilot to scale fast enough once it works. “There’s a massive difference between a use case that works in an isolated environment and one that reshapes a function.”

Most hotel groups have a hundred isolated environments. That’s called a portfolio.

The trust ladder — and the fact that two very different people built the same one

Here’s the convergence that makes this worth an article rather than a LinkedIn post.

BCG, advising CIOs of global enterprises, prescribes graduated autonomy as the governance model for agents that take real action: “shadow mode, supervised mode, guided autonomy, full autonomy. Each tier is earned through demonstrated performance.”

Mews, building agents inside a hotel PMS, arrived at the identical structure and gave it a better name. Rogers Leo, Mews’ Director of Engineering, calls it earned autonomy:

“Every agent Mews builds starts with human-in-the-loop oversight. Agents propose. Humans approve. This isn’t a temporary training wheel. It’s a design principle… You don’t start by handing decisions to agents. You start by having them show their working.”

A global consultancy and a hotel-tech engineering team, working from completely different vantage points, independently landed on the same answer. When that happens, it’s usually not fashion. It’s structure.

And Leo’s version is sharper for our industry, because he grounds it in what actually goes wrong on a Tuesday: “A wrong rate published across channels costs real revenue. A VIP’s preferences ignored costs a relationship. A maintenance issue left unrouted costs a guest’s experience. These aren’t edge cases. They’re Tuesday.”

He adds three things BCG’s enterprise framing misses, and all three are hospitality-specific:

Decision transparency. Staff trust doesn’t come from understanding the model. It comes from seeing the reasoning. “When a front desk agent can see that a room was suggested because it matched a guest’s historical preferences and avoided a housekeeping conflict, they can evaluate, challenge, or confirm that recommendation with confidence.” That’s what converts an agent from opaque automation imposed from above into something that fits inside an existing SOP.

Closed-loop learning. Every override is training data. “A property that has been running an agent for six months has a fundamentally different — and better — system than one that just turned it on.” This is the compounding argument for starting now, and it’s the one that should worry anyone still waiting for the technology to settle. The front desk manager who knows Mr. Chen wants the corner room on six isn’t overridden by the algorithm — his knowledge gets encoded into it.

Tone, taste, and brand. The dimension no enterprise AI framework accounts for: “An agent that produces a correct outcome in the wrong voice — too casual for a luxury brand, too formal for a lifestyle property — has still failed.” Functional accuracy is table stakes. In hospitality, the evaluation bar is “did it get the answer right in a way that reflects who we are as a brand?”

So what do you actually do

Point it at operations first, not the guest. Ben Rafter, CEO of Hotel Equities, has been blunt about this: AI is delivering provable returns fastest on the operational side — scheduling, procurement, revenue management, maintenance triage — while guest-facing chatbots and in-room assistants remain comparatively immature and harder to prove out. Labor is the largest controllable line on the P&L, and what managers need there isn’t more data, it’s recommendations precise enough that they’ll actually act on them. Unglamorous, and it’s where the money is. Awad’s filter applies cleanly: “If a process is simple and rule-based, you don’t need an agent.” Agents earn their keep in complexity — multiparty information exchange where interpretation is needed. That describes a hotel almost perfectly.

Change the scoreboard from adoption to value. This is BCG’s second CEO imperative and it’s the one hospitality is failing hardest. “Adoption tells you that people use AI, not whether it pays off.” You already know 74% of your people use it. That number is now worthless to you. The question is what happened to the eight hours. Rafter’s questions for any management company pitching AI are the right ones for your own initiatives: which KPIs moved, over what period, against which baseline, and are the gains durable?

Decide what the saved day is for — and say it out loud. Sixty-six percent of employees get no guidance on this. In hospitality that ambiguity isn’t neutral, it’s frightening: staff assume the answer is headcount. Kingsmill and Reti’s IKEA example is the alternative worth stealing — when the chatbot absorbed routine queries, IKEA reskilled those workers into a paid interior-design consultation service, converting displacement into a new revenue line. The question is not what regulation permits. It’s whether you redeploy the hour toward the guest or just delete it.

Build the ladder before you build the agent. Write down your four rungs — shadow, supervised, guided, autonomous — and the measured performance that earns promotion between them, before anything goes live. Then hold to it. This is what turns “we’re piloting AI” into something a GM, a brand standard, and an owner can all live with.

Co-create with your best people, not your vendors. Abraham’s separator between the 5% and the 95% is exactly this: “The companies at the forefront build their agentic solutions with their best people around their real day-to-day challenges.” Your best front desk manager and your best revenue manager should be shaping the agent weekly. Not receiving it at rollout. Schulze gave every employee $2,000 of authority to fix a guest problem because he trusted their judgment. The modern version of that trust is letting them define what the agent should do.

The uncomfortable close

61% of BCG’s respondents believe that within three years, AI agents could do at least half their job. Not the vendors. Not the analysts. The people doing the work.

Whatever you think of that forecast, your staff have already made it. They’re using the tools daily, finding a day a week, and watching leadership say things that only 28% of them believe connect to what the organization actually does.

The industry spent two years asking whether hospitality would adopt AI. It has. The question now is whether hotels will redesign anything around it — or whether the biggest productivity gain the industry has seen in a generation quietly evaporates, eight hours at a time, because nobody ever said what it was for.

---

Sources: BCG, The Agentic Leadership Playbook (July 2026) and AI at Work: Why Strategy Matters More Than Tools (June 2026); Mews, Building Trust: How Agentic AI Earns Its Place in Hospitality (June 2026); plus reporting from Hospitality Net, Hospitality Daily, and the AI Hospitality Alliance.