n8n Gmail Triage Workflow: GPT-5.6 Email Labeling

Automatically triage your unread Gmail with GPT-based labeling—no manual inbox sorting

This n8n workflow polls your Gmail every minute, sends each unread email to an OpenAI Chat Model for strict JSON classification, and then automatically applies the right Gmail label (or marks it as SPAM) to keep your inbox clean, professional, and prioritized.

What this workflow does

• Checks Gmail every minute for unread messages.
• Extracts key fields from each email: message ID, sender, subject, and plain-text body.
• Classifies with OpenAI by sending the email content to an OpenAI Chat Model with instructions to return a strict JSON result (category, priority, summary, and an action flag).
• Parses and validates the AI response, using a safe default classification if JSON parsing fails.
• Routes by chosen category and applies the matching Gmail label.
• Protects your inbox by applying Gmail’s SPAM label when the action flag indicates it.

Use cases

• SaaS operators who want unread emails categorized into Client, Prospect, Administration, Facture, Newsletter, and Urgent without manual work.
• Automation engineers testing reliable AI-driven email triage with structured JSON outputs and fallback behavior.
• Teams managing high email volume who need consistent priority labeling to reduce response time.

Technical details

• n8n Gmail integration using a Gmail Trigger plus Gmail actions to add labels or apply SPAM.
• OpenAI Chat Model for GPT-based classification and summarization (via strict JSON format).
• Logic and orchestration nodes: if, set, code, gmail, merge, switch.
• Setup requirements:
  • Add a Gmail OAuth credential for the Gmail Trigger and Gmail actions.
  • Add an OpenAI credential for the OpenAI Chat Model node.
  • Create Gmail labels: Urgent, Client, Prospect, Facture, Administration, Newsletter, then copy their label IDs and replace each YOUR_GMAIL_LABEL_ID_... placeholder.