Triage Slack and Gmail requests with an AI-powered intake layer

This n8n template helps you turn inbound messages into a clean, deduped queue of actionable tickets.

It includes Slack and Gmail as ready to use examples, but the key idea is the universal intake normalizer: you can plug in other sources later (forms, webhooks, chat tools, other inboxes) as long as you map them into the same normalized schema.

Good to know

• This workflow sends message content to an LLM for classification.
• Keep sensitive data out of the prompt, and only process messages you are allowed to process.
• Costs depend on message volume and length, so truncation and tight filters matter.

How it works

• Collect inbound items (Slack and Gmail are included as examples).
• Normalize each item into one shared JSON format so every source behaves the same.
• Deduplicate items using a data table so repeats are skipped.
• Use an AI agent with structured output to score urgency and importance, produce a summary, and draft a reply.
• Create a Notion ticket for tracking, and optionally notify Slack for high priority items.

Setup steps

• Connect credentials for Slack, Gmail, Notion, and your LLM provider.
• Choose your Slack channel and set a Gmail filter that keeps volume manageable.
• Select your Notion database and ensure properties match the field mappings.
• Create or select a data table and map the unique ID column for deduplication.
• Adjust the notification threshold and schedule interval to match your workflow.

Requirements

• Slack workspace access (optional if you swap the source)
• Gmail access (optional if you swap the source)
• Notion database for ticket creation
• LLM API credentials

Customising this workflow

• Add new sources by mapping them into the normalizer schema.
• Truncate long messages before the AI step to reduce cost.
• Change categories, scoring, and thresholds to match your operating model.