Hospitality & Leisure

Through the interfaces your tools already expose, without replacing them. The analysis layer runs inside your perimeter or in sovereign cloud, and falls back to your business rules if it becomes unavailable: a point of sale does not stop because a model stopped answering. The real preparatory work is elsewhere, reconciling your product, slot and site reference data, which rarely carries the same identifiers across reservation, till and accounting.

A clarification first: NIS 2 applies to you, the regulated entity, not to us, and being in scope is checked property by property rather than at sector level. Where you are in scope, we separate point-of-sale networks from the rest of the estate, produce the incident register and risk analysis the regulator expects, and replay the continuity plan. On payment flows, the card security standard adds its own requirements, and those are held continuously rather than ahead of an audit.

Three documented projects. At Disneyland Paris, video flow analysis applied to ride access cut waiting by 58% and lifted satisfaction by 26 points. At Compass Group, visual tray recognition cut time at the till by 82% and keying errors by 36%. At B&B Hotels, introducing agents into the development chain doubled delivery velocity. We would rather offer those three cases than a sector average: your footfall and your starting point will decide your own figures.

A recommendation system falls under limited risk in the European AI regulation, not high risk: the obligations are about transparency, not certification. In practice, a guest interacting with a machine must know it, and be able to reach a human. The real point of attention is elsewhere: recommendations that consistently push the same segment towards the most expensive offers show up in the data before they show up in reputation. We monitor recommendation diversity as closely as performance.

Framing takes 2 to 6 weeks depending on scope and the number of systems to connect. A first version tested at one point of sale or one attraction follows in 4 to 10 weeks, depending on the quality of your footfall data. A multi-site rollout, including training the operations teams, runs to 3 to 6 months. These ranges are the ones we apply everywhere, and each comes with what makes it vary.

Per stay and per cover, across the three items that actually weigh: guest travel, food, and site energy. An average per room night is useless for deciding, because it hides exactly the variation you can act on. On regulatory reporting, the Omnibus directive of 18 March 2026 raised the thresholds to one thousand employees and four hundred and fifty million euros of revenue: the first question is whether you are still in scope, and many operators no longer are.