Gmail AI Labeling with TypeSafe Jev (System One) | n8n

Automatically label your unlabelled Gmail messages with TypeSafe Jev (System One) — safely

This n8n workflow checks your Gmail every five minutes for unlabelled emails, classifies them using TypeSafe Jev (System One), creates any missing Gmail labels, and applies the best label—while routing uncertain messages to a safe “Needs Review” fallback.

What this workflow does

• Scheduled search (every 5 minutes): Looks for up to 50 Gmail messages that have no user labels, excluding chats, sent mail, and drafts.
• AI classification: Sends each email’s headers, snippet/body, and Gmail category signals to TypeSafe Jev (System One) to return:
  • a single label choice
  • a confidence score
  • a “needs reply” score
• Confidence-based routing: Applies confidence thresholds to choose the final label name, escalating uncertain-but-actionable messages to “Action Required” and sending low-confidence results to “Needs Review”.
• Label management: Compares the chosen label names with existing Gmail labels, creates missing labels once to avoid duplicates, then reloads labels to obtain fresh label IDs.
• Apply labels in Gmail: Resolves the selected label name to a label ID and applies the label to each message.
• Drain the backlog: Repeats the search-and-label cycle within the same execution until the backlog is drained or a maximum rounds limit is reached.

Use cases

• Keep an inbox organized by auto-labeling messages that arrive without manual tagging.
• Route potentially time-sensitive emails to Action Required while protecting against mistakes with Needs Review.
• SaaS operations teams that want consistent triage for support- or sales-related inbound email.

Technical details

• Nodes/blocks: schedule trigger, if, code, gmail, sticky note, http request.
• TypeSafe integration: HTTP Request to TypeSafe API using TypeSafe Jev (System One) with a Bearer API key (via an HTTP Bearer Auth credential).
• Gmail integration: Gmail OAuth2 credential with permissions to read messages and manage labels.
• Setup step: Review and adjust the confidence thresholds used to pick labels.