PLATFORM
Controls & governance
Human-in-the-loop, brand voice, phased deployment, audit logs.
Configure per-channel autonomy, confidence thresholds, and property-type guardrails. Full conversation history with agent rationale is queryable for review.
Unified Inbox Translation
Guests write in 100+ languages. The unified inbox translates a chat into the preferred language of the person reading it, so one teammate can follow in English and another in German. The translation is for the team. The guest still reads and receives messages in their own language. Product view: Messaging.
Phishing guardrail
Operators can turn on a phishing guardrail so the AI recognizes payment links and communication channels that come from a potentially harmful phishing attempt, on messaging, voice, and email. Guests are contacting properties more often to ask for help and clarification. The guardrail is a preventive step for that spike. Detail: Agent safety.
