Somewhere in the day's 120 arrivals (the Radicati average for a working professional) are the handful that will matter in a month: the client tilting toward leaving, the approval unblocking a launch, the candidate about to accept elsewhere. The other hundred-plus are receipts, reflexive ccs, newsletters, and threads that would resolve identically without you. The entire skill of modern email is telling these apart quickly, and it reduces to three checks.
The three signals
- A direct ask, aimed at you. Not a thread you're copied on, not an FYI: a question or request where you specifically are the answerer. If you can't find the ask in ten seconds, there usually isn't one.
- A date or decision attached. Deadlines, meetings, approvals, expiring offers: anything where time converts inaction into an outcome. Dated items are automatic extractions to the task list, whatever else is true of them.
- A sender whose mail has history. Some correspondents' messages always need action (your manager, live clients, the notary); most senders' never do. Your own reply history is the database, and you already know it intuitively.
Score each message against the three in seconds: none present, archive on sight (this is most of the feed); one or more, it's real: answer it, extract it, or log what it makes you wait for.
What percentage of email actually needs you?
Less than feels safe to admit. Studies of workplace email find only a minority of received messages ever get replies, and the share carrying genuine decisions is smaller still; for most professionals the honest ratio is single digits out of the daily 120. The feed's visual uniformity hides this ruthlessly, which is why the important ones get missed: they look exactly like everything else.
Teaching the check to a machine
The three signals are also, not coincidentally, what AI triage models read: content (is there an ask, a date), relationship (does this sender's mail historically need you), and thread state (whose court is the ball in). Run continuously, the check becomes a standing filter: Pidgy sorts a Gmail or Outlook inbox into needs-you, waiting-on, and reference, so the ten that matter sit on top when you open it, already separated from the hundred that don't. The mechanism in full: how AI triage decides.