The real bottleneck for AI in hotels isn't the technology — it's training

Two articles published this summer in Lodging Magazine, the official journal of the American Hotel & Lodging Association, tell a story together that's worth bringing to Italy. The first, by Shruti Shibulal of Tamara Leisure Experiences, is about leadership and technological transformation. The second, an interview with Priya Rajamani of StayNTouch, gets into the practical detail of how you actually train staff on a new property management system. Read together, they sketch out a problem that concerns every property, large or small: technology in hotels almost never fails because of the tool's limitations — it fails because of how it's introduced to the people who have to use it every day.

The problem isn't where to put the technology, but what you're asking of the people who'll use it

Shibulal opens with a statistic I consider the heart of the whole piece: according to a 2025 survey by the American Hotel & Lodging Association and Hireology, 65% of American hotels report staff shortages, and 71% have open positions they can't fill. In this context, she writes, a badly implemented technology doesn't solve the staffing problem — it makes it worse: it increases operational complexity at precisely the moment when teams are already under pressure. The right question to ask, she argues, isn't where you can apply artificial intelligence, but where it can create the clarity, capability and confidence people need to do their jobs well.

The other point I find valuable is the distinction between automation and capability augmentation, which the article refuses to treat as a binary choice between human service and machine. A revenue manager can benefit from an AI that spots anomalous demand patterns, but it's still up to them to interpret the commercial meaning. A front-desk agent can receive an alert about a room that isn't ready on time, but they're the one who decides how to communicate that to the arriving guest. In every case, the technology supports the decision — it doesn't take full responsibility for it. It's a distinction I consider fundamental even for a small property: responsibility is never fully delegated to an algorithm. You always have to establish clearly what to automate, what to augment, and what to leave deliberately in human hands.

How you actually train a team — not just in theory

If the first article explains the why, the Rajamani interview explains the how — with a very concrete detail that I think is the most common mistake, and one I recognise perfectly well in our own market: treating training as a one-off event, a session before the system goes live, then leaving new hires to learn from colleagues who may not remember every detail themselves. Rajamani proposes a different model: role-specific learning paths rather than one generic training for everyone, a protected practice environment where people can experiment before touching a real booking, comprehension checks throughout the process and not just at the end, and visibility for managers over each team member's progress — so they can step in before a training gap becomes an operational problem visible to the guest.

One passage in particular speaks directly to anyone managing an independent property like mine: Rajamani observes that integrations between different systems matter most for independent and boutique hotels — those that aim to distinguish themselves precisely through a particular service or offer — but they also remain the primary source of confusion for staff when a property is juggling ten or more different vendors. The practical advice she draws from this is to always have, within your own management tool, direct access to vendor documentation, so that staff can understand not just what each integration does but also how it communicates with the rest of the system, without having to call support for every question. It's the same theme of genuine system openness I raised when reporting on the closed-door conversation between the CEOs of the major PMS providers and the CIOs of the large chains.

The Italian context makes this problem even more urgent

In Italy the picture is, if anything, more fraught than the American one. Demand for seasonal staff in the hotel and restaurant sectors is forecast to grow 12% in 2026, while the attrition rate in the industry had already reached around 20% in 2024. According to Unioncamere's (the Italian Union of Chambers of Commerce) analysis, the shortage of qualified staff risks holding back growth and innovation in the sector, hitting small and medium-sized businesses especially hard — and they make up the vast majority of the Italian hotel landscape. In this scenario, training every new person who joins the team quickly and well — whether they're seasonal or permanent — stops being an organisational detail and becomes a genuine competitive lever, perhaps more urgently so in Italy than in the United States.

What an Italian operator can actually do

The practical takeaway I bring home from these two articles, for anyone running a small or medium-sized property, is this: never separate the choice of a new technological tool from the question of how it will actually be learned by the people who'll use it every day. Before adopting any new system, it's worth asking clearly what it should automate entirely, what it should only support while leaving the decision to a person, and which interactions with guests should deliberately remain human. It's also worth applying the same rigour I described in the piece on the vendor evaluation framework — designed for an enterprise CIO and adapted to our scale. And it's worth building, even in a small property, a training path that doesn't exhaust itself on day one, but accompanies every team member with progressive checkpoints — not just a final test that's forgotten immediately after.

This is exactly the kind of work I'm carrying forward outside my own hotel too, as a trainer for the sector, with courses on AI applied to hospitality for trade associations and tourism destinations. If you're thinking about how to introduce new AI tools in your property without wasting energy and without leaving your team alone in the face of change, I'm available for personalised training and consultancy on this specific topic, built around the operational reality of each individual property — not a one-size-fits-all theoretical model.

FAQ

Why does AI adoption in hotels fail more often because of training than because of the technology? Because a technology introduced without an adequate training path increases operational complexity instead of reducing it, compounding problems that already exist — like staff shortages, which according to an AHLA-Hireology survey affect 65% of American hotels.

What distinguishes automation from capability augmentation in a hotel? Automation replaces a task entirely, while capability augmentation supports a decision that still rests with a person — for example, an AI alert about a delayed room that the staff member decides how to communicate to the guest.

What is the most common mistake in training staff on a new property management system? Treating training as a one-off event before the system goes live, rather than building a continuous path with progressive checkpoints, protected practice environments and role-specific learning tracks.

*cover image created with AI