Passive sensors
Connect once. Listen forever.
Surveys ask people how they feel. Integrations show you how they behave. We read metadata only — who meets whom, when, how often. Never meeting contents. Never message bodies.
📅
Google Calendar
Structural signals · 1:1 cadence, meeting load, deep-work erosion
○ Not connected
What we detect
- →1:1 cancellation streaks — leading indicator of resignation 90 days out
- →Meeting overload — >15 hrs/week of meetings predicts burnout
- →Deep-work erosion — longest uninterrupted block <90 min
- →After-hours creep — meetings before 8am or after 7pm
- →Network shrinkage — unique collaborators dropping month-over-month
We read event times, attendee count and response status — never titles, descriptions or attachments unless you opt in.
Live signals · last 30 days
Reading your calendar…
💬
Slack
Behavioral signals · response latency, channel withdrawal, reaction sentiment
What we'd detect
- →Response latency drift — DM reply time tripling = early disengagement
- →Channel withdrawal — stopping posts in #team-banter happens ~60 days before quitting
- →Reaction sentiment shift — ❌/😡 ratios climbing in team channels
- →After-hours DMs from manager — burnout transmission vector
- →Decision velocity — threads marked "decision needed" → time to ✅
Metadata only. Never message contents in v1 — that's the line we don't cross.
Slack ships with the Org / Enterprise tier alongside cross-team rollup, SSO and HR exports.
See enterprise tierPrivacy contract: managers see "team response latency: amber" — never "Sarah took 4h to reply to Jake". This is what gets HR & Legal to sign instead of block.