HITEC Paris: Own Your Hotel Data Layer Before Adding AI Agents
At the inaugural HITEC Paris, Belmond's Fabio Maggioni argued hotel companies should own their data layer first, since adding AI to fragmented systems only raises costs.
Hotel companies should own their data layer before they add AI agents, Fabio Maggioni of Belmond told the AI panel at the inaugural HITEC Paris, because adding AI to fragmented systems only raises costs. Hospitality Net’s report from the panel, contributed by Travel Singularity, sets out the argument.
HITEC Paris ran alongside EquipHotel, organised by RX Global with HFTP. Moderator Lyle Worthington was joined by Maggioni, Alexander Schmidt of Hotelschool The Hague and Simone Puorto. Schmidt’s research found hospitality is among the slowest sectors to adopt AI. He warned that AI assistants could become a new intermediary layer, like online travel agencies, eroding a hotel’s control of the guest relationship, and proposed judging AI investments on operational efficiency, employee experience and guest experience. Maggioni’s case was practical: with an owned data layer, a company can connect third-party systems and build its own agents.
The panel also raised the need for a single source of truth as AI agents become the guests, dependence on a few foundation-model providers, and the risk that automation removes the entry-level roles that develop future leaders.
The adoption gap is already visible in the numbers. A Hospitality Net brief reports an AHLA and Actabl finding that 84% of respondents said AI improved their own work, compared with 15% for frontline work.
For an owner or asset manager, the operating question is where the data lives. If a vendor reads from your systems of record, you can swap the vendor. If you pay a vendor to hold the only clean copy, you are renting your guest relationships, and the cost per guest rises with every new tool.
Source: Hospitality Net — HITEC Paris 2026: The Rise of AI Agents and the Value of Being Human Auto-generated brief — verified before publishing.