Early burnout-risk signals for HR from anonymised work patterns — meeting overload, after-hours creep, weekend work — without reading a single message.
Workplace burnout costs an estimated $1 trillion a year, and HR usually finds out last. This agent gives you early risk signals from aggregate work-pattern METADATA only — meeting density, calendar overload, after-hours creep, weekend work, daily span — ingested from Google Calendar (read-only) or CSV exports. It never reads message or event content. People are stored under pseudonymous keys, team summaries respect a k-anonymity minimum, and the 0–100 score is deterministic with explicit drivers — not a black box. Signals are indicators, not diagnoses: built for supportive conversations, not surveillance.