Risk-Tiered AI Governance Engine

Policy Controls, Risk Thresholds & Brand Tone Governance

Active Policy: v1.1

Tier 1: Low Risk (0 - 25)

Auto-Dispatched

Wifi passcodes, check-in time inquiries, parking instructions.

Tier 2: Medium Risk (26 - 50)

Human Queue Optional

Late checkout inquiries without extra cleaning fee charges.

Tier 3: High Risk (51 - 75)

Human Approval Mandatory

Maintenance complaints, noise reports, amenity refunds.

Tier 4: Critical Risk (76 - 100)

Hard Lock / Blocked

Cancellation requests, dispute threats, direct bank transfers.

Threshold & Risk Score Settings

Auto-Reply Maximum Risk Score:25.0 / 100

Prompts scoring above 25.0 will be routed to the Human Approval Queue.

Mandatory Human Review for Financial Requests

Forces review for refunds, credit cards, or payouts

Mandatory Human Review for Cancellation Requests

Prevents AI from automatically cancelling active reservations

PII Masking & Redaction Gateway

Masks credit cards, emails, and lock PINs before LLM call

AI Persona & Brand Tone Selection

Warm Professional (Default)

Hospitable, clear, polite, and reassuring with precise property details.

Formal Luxury Concierge

High-end luxury tone tailored for 5-star boutique estates and penthouse suites.

Casual & Friendly

Relaxed, modern, and approachable tone suitable for urban lofts and vacation beach houses.