Security & data
Guest memory is the product — so protecting it is the architecture. The AI never receives unrestricted access to hotel systems or physical devices; every action passes through authentication, authorization, permissions and safety rules.
Encryption in transit
TLS 1.3 for every guest, staff and device-facing channel.
Encryption at rest
Preference and profile data encrypted with managed keys.
Role-based access control
Staff roles scoped by property, department and action.
Authentication
SSO / SAML, MFA and step-up authentication for sensitive actions.
Authorization
Every AI action evaluated against a policy before execution.
API security
Scoped keys, rotation, rate limits and signed webhooks.
Audit logs
Immutable record of intent, decision, action and outcome.
Data minimization
Only preference fields required to configure the stay.
Tenant isolation
Hard isolation between hotel groups and properties.
Credential management
Vendor credentials stored in a managed secret store.
Permission-based device control
Device commands bound to guest and room scope.
Retention & consent
Guest-controlled retention, export and deletion.
Data path
Guest data
Consented preference records, scoped to the guest.
↓
Secure data layer
Encrypted storage, tenant isolation, retention policy.
↓
Preference engine
Resolves the profile into a target room state.
↓
AI orchestration
Plans, classifies risk and requests permission.
↓
Authorized hotel systems
Executes only through approved adapters and scopes.
Governance
audit_id aud_demo_8f21c tenant meridian-house actor ai.concierge.v3 guest gst_demo_2104 (consent: granted) intent climate.set_temperature risk LOW policy room.climate.write @ stay-scope decision ALLOW adapter hvac.generic.v2 result SUCCESS (142ms) retention stay + 90 days
Demonstration log entry.