Gmail AI Email Classification with GPT-4o-mini Labels

Turn new Gmail emails into labeled, actionable inboxes—automatically

This n8n workflow watches your Gmail for new messages, uses OpenAI (GPT-4o-mini) to classify each email, then automatically applies the matching `Gmail label` based on the predicted category—so your inbox stays organized without manual triage.

What this workflow does

Designed for practical inbox automation, this “lite” build follows a simple pipeline:

• Triggers every minute when a new email arrives in Gmail.
• Builds the OpenAI request using a classification prompt, your selected model (GPT-4o-mini), and a mapping of categories to Gmail label IDs.
• Sends key email details to the OpenAI Chat Completions API: the email subject, sender, and a truncated body.
• Parses the JSON response to extract category, urgency, and a short summary, defaulting to “noise” if parsing fails.
• Adds the corresponding Gmail label to the original message based on the returned category.

Use cases

• Automatically sort inbound leads, support emails, and newsletters into labeled Gmail folders.
• Reduce inbox backlog by tagging “noise” or low-priority messages without opening them.
• SaaS operators who want consistent triage rules without maintaining complex filters.

Technical details

• n8n nodes: Gmail Trigger, Set, HTTP Request, Code, and a Gmail label action (plus notes via Sticky Note).
• Integrations: Gmail (trigger + apply labels) and OpenAI (Chat Completions) using HTTP Header Auth.
• Setup requirements: connect Gmail credentials, add your OpenAI API key, create Gmail labels, copy their label IDs, and replace the placeholder values in the category-to-label mapping.

Note: This is the lite version. A fuller build also adds a Google Sheets log, AI-drafted replies to new enquiries, and instant alerts for urgent emails (available at baynhams.gumroad.com/l/inbox-triage).