AI transparency

The platform offers an optional AI concierge and AI-assisted content tools, enabled per hotel. This page explains how AI use is disclosed to guests and kept safe and grounded. It describes our transparency posture; it does not claim "EU AI Act compliance."

Guests know when they are using AI

• Where the AI concierge is enabled, guests are told they are interacting with an AI assistant — not a human agent — and the disclosure is shown consistently on the AI surface.
• A guest can always ask for a person. A human handoff / escalation path exists or is clearly described, and serious or sensitive requests are routed to hotel staff.

Grounded in the hotel's facts

• AI answers are grounded in hotel-provided information and live offerings. The concierge books only from the hotel's live catalogue, and prices, availability, and booking details are confirmed by the hotel.
• AI output is not presented as legal, medical, emergency, or official hotel policy unless it is sourced from approved hotel content.

Safety-critical content

• Allergen and dietary safety information is never auto-published in a translated form without a human review gate per language. It is treated as a structured safety fact, not free text routed through the AI translation cache.
• Structured facts (prices, availability, tax, identifiers) are never routed through the translation or AI caching path.

Data and sub-processors

• AI inference uses Amazon Bedrock in AWS us-east-1 (United States) (Claude and Titan models). The prompt context sent for inference is minimised and excludes stored payment card numbers.
• Conversation logs and their retention follow the hotel's privacy and retention settings. See the Data Map, Privacy, and Sub-processors pages.