AI in hospitality, part 1 of 15
A consultant visits the property and talks about AI assistants, automated workflows, perhaps a digital concierge. Everyone nods. Three months later either nothing is running at all — or a single chatbot that helps nobody in particular.
Usually there is no great disaster behind this. Just a gap that nobody looked at beforehand.
What are data silos in a hotel operation?
Data silos arise when guest information is spread across several systems that are not connected to each other — for example PMS, point of sale, accounting, HR software and CRM. Each system stores only a fragment of the information about the same guest, without the other systems knowing about it.
Five systems, five truths
A typical property works with a PMS for room management, a point-of-sale system for restaurant and bar, accounting software, an HR solution for rosters and payroll, plus a CRM or an Excel list for regular guests. Five systems. They rarely talk to each other.
That is not a problem in itself. The problem arises because each system knows only a fragment of the same guest. The PMS knows that Mrs Meier prefers a room with a balcony. The CRM knows that she is booking for the fourth time. The kitchen knows that she eats gluten-free. None of the three knows what the other two know.
This is not an isolated case: according to the 2026 Hotel PMS Impact Study by HotelTechReport, 45 per cent of the hoteliers surveyed name faster integrations and open interfaces as the most urgent development for their PMS. And according to the HEDNA State of Distribution Report, 67 per cent of independent hotels describe dealing with separate systems as one of their biggest operational challenges.
Why this becomes a problem for AI
An AI assistant that pre-sorts emails, answers enquiries or forecasts occupancy sees only the data it is given. If that data is scattered across five systems, it sees at best half the picture. It answers an email correctly without knowing that the same guest has just filed a complaint about the room. It suggests a room that housekeeping has not yet released.
The obvious reaction: "The AI doesn't work." In fact it did exactly what was possible with the available data. The real problem sits one level below.
First the tool, then the disillusionment
A common pattern: a property hears about a promising AI tool, introduces it — and wonders why the effect fails to appear. Rarely is the tool to blame. Usually it is the order of steps.
Before an assistant, an automated workflow or an autonomous agent can work usefully, some questions need answers: which data exists? Where does it sit? Who is allowed to access what? And, not to be underestimated where data protection is concerned: what is legally permissible at all? That is less spectacular than a tool demo. But precisely this work decides later between success and frustration.
An exercise for your own operation
Follow the path of a single guest through your systems — from the booking enquiry to the invoice. At how many points does somebody transfer information by hand from one system to another? And how often does a preference, a note or a special arrangement get lost along the way?
Each of these points is a seam. If it stays open, every later AI deployment runs into the void there. A structured assessment therefore does not start with the question of the tool, but with the question of the foundation.
How the series continues
The coming articles deal with AI in hospitality — not only in marketing, but also in staff planning, accounting, at reception, in revenue management and in internal communication. What connects all of them: a clean, consolidated data foundation decides whether AI can have any effect in a given area at all.
The next part is about how to connect systems sensibly without overhauling the entire IT landscape.
Further reading
- Agentic AI: why hotels need to rethink their booking systems now
- Why don't hotels have their own AI budget — and what does that really cost?
- AI in the hotel: start now or wait for the next generation?
Sources
- HotelTechReport (2026): 2026 Hotel PMS Report: What Elite Operators Want, Use, and Plan Next
- HEDNA State of Distribution Report, cited in Cloudbeds (2026): PMS Integrations: What They Mean for Hotels in 2026
Frequently asked questions
Why do AI projects in hotels often fail?
Usually not because of the chosen tool, but because the underlying data is spread across several separate systems (PMS, CRM, accounting, HR) and nobody consolidates it. An AI assistant can only work with the data it is allowed to see.
What is the first step before introducing AI in a hotel?
A structured assessment of systems, data, team and processes — before a tool is selected. Tourismusconsult offers the AI readiness assessment for this, which results in a prioritised roadmap across three time horizons.
Which systems in hotels are typically separated from one another?
Most commonly: property management system (PMS), point-of-sale system, accounting software, HR solution for staff planning and CRM for guest relationships.
Tourismusconsult accompanies hotel operations through the AI readiness assessment: a structured analysis of systems, data, team and processes that leads to a prioritised roadmap — with quick wins, medium-term and strategic measures.
