Hours, overtime, short turns, consecutive days, checkout load, arrival waves, training readiness, manager coverage, and role clarity.
People protection layer
The system may notice pressure. It may not pretend that pressure is a person.
A hotel can hit budget and still burn through the team. The people layer catches operating strain early enough for support: extra coverage, manager attention, room-board changes, training refreshers, recovery scripts, and workload adjustments.
Voice, fairness, recognition, support, and role clarity can inform the read only when response thresholds protect individuals.
No individual performance ranking, discipline, termination, health inference, loyalty inference, intent inference, or protected-class inference.
Buyer-safe cockpit preview
Change the pressure inputs and see the support route.
This public cockpit demonstrates behavior, not protected scoring weights. Production calculation logic should run server-side behind account access.
Workflow integration
People protection is not separate from financial performance.
Evidence and compliance posture
Trust requires worker-centered design and risk controls.
NIST frames AI risk management around impacts to individuals, organizations, and society; Hotel OS translates that into visible guardrails and reviewed use.
DOL worker AI principles emphasize transparency, worker engagement, rights protection, and using technology to enhance work.
EEOC guidance warns that employment tests and selection procedures can create discrimination risk. Hotel OS blocks use of people pressure signals as selection or discipline logic.
Open EEOC Employment Tests and Selection Procedures guidance