Hospitality rarely has one central AI system that affects the whole business. Instead, there are separate applications: a dynamic pricing system for rooms or tables, a chatbot that handles reservations, a tool that builds schedules based on expected busyness, software that summarizes reviews or responds to guests. Each of these applications is limited on its own. Together they form a layer of decisions that no one has recorded centrally.
On top of that, hospitality often works with a small permanent core and a larger flexible layer: seasonal staff, temporary workers, multiple locations with their own manager. Whoever introduces a tool at one location does not automatically share that with the others. The ratio between what is known centrally and what is actually used on the floor is therefore usually more skewed than in sectors with a tighter organizational structure.
The IT department, if there is one, manages the point-of-sale systems, the booking platform and perhaps the scheduling program. What the location manager has purchased or activated themselves — an AI feature in the booking tool, a separate subscription for responding to guest reviews, an app that suggests staff planning — rarely appears there. This is not negligence; it is the result of how decisions are made in this sector: decentralized, at the location, by people who want to see results and do not wait for an approval process.
The consequence is that an overview that starts with the IT list gives a distorted picture. The question is not only what has been purchased, but what is being used: by the manager who adjusts prices in the evening using a separate subscription, by the front desk that lets a chatbot respond to guests, by the planner who has a tool weigh in on staffing. That usage becomes visible by asking, not by searching in a procurement system.
If that question is asked as a form of control — who introduced something here without permission — you will not get a complete answer. People who purchased something because it worked will not report it if the first reaction is a correction. The inventory only works if the starting point is: what is running, regardless of how it got there. That is a different attitude than an audit, and that attitude is needed to bring shadow AI into view. How to have that conversation in a way that makes people willing to tell you what they use is described on the page about how to prevent employees from using a tool no one knows about.
A pricing tool that adjusts room rates, a chatbot that communicates with guests, and a scheduling tool that assigns staff are different in nature. One system drives revenue, another directly affects the customer relationship, the third affects working conditions. They do not belong in the same category. A governance approach that works in hospitality looks at what each system does and for whom, and attaches a risk level to that — not to the name of the supplier or the price of the license.
That classification differs by sector. In financial services, the emphasis lies on systems that affect credit decisions, as described on the page about AI governance in financial services. In the real estate sector, it is more often about valuation and appraisal models, as described on the page about AI governance in the real estate sector. Hospitality has its own mix: guest contact, dynamic pricing and staff planning, each with its own risk profile.
A hospitality business usually already has some form of risk management: food safety, working conditions, fire safety. AI governance does not need to be a separate structure alongside that. The question is whether an AI application that sets prices, talks to guests or schedules staff fits into the same structure as other operational risks. Often it does — it then comes down to who decides, who provides oversight and how that is recorded, not a new compliance process.
This approach recurs in other sectors with a lot of decentralized decision-making and little central IT management, as described on the pages about AI governance in the agricultural sector and about AI governance in the energy sector. A similar issue also arises in the ICT sector, where a lot of technical knowledge is present but tooling is expanding rapidly, as described on the page about AI governance in the ICT sector.
Once it is clear which AI applications are actually running in a hospitality business and what role they play, a follow-up question arises that is not about risk but about how the work is organized: which part of a task — adjusting prices, answering guest questions, drawing up schedules — can a system take over, and which part remains human work. That question is answered by the work scan from FTE TO AI, which calculates per task which part qualifies for AI, separate from the question of how that usage is governed.
Vraag maar. Governance begint bij weten wat er draait — ook wat niemand heeft goedgekeurd.
Answers come from this site’s knowledge base. Not tailored advice, and not a scan of your company.