agentusecasesAll 965 use cases
Sales, Marketing & Support

Triage support emails and draft replies for human review

An n8n workflow classifies, prioritizes and summarizes each support email, drafts a reply and posts it to Slack for a person to send.

Done withn8n

What they did
A five-node n8n workflow polls Gmail for new support emails. Each one goes to an OpenAI-compatible model (gpt-4o-mini in the shared JSON), which returns a category, an urgency, a one-sentence summary and a 3-5 sentence draft reply. The results are logged to a Google Sheet and posted to Slack. A person reviews the draft and sends it, since the author kept sending manual on purpose.
How it went
The author says it takes about 20 minutes to wire up and that the drafts are good. No volume or time-saved figures are given. Commenters found bugs: one email per poll is processed, and the sender shows as an object.
Worth knowing
Before production, fix the Code node so it handles every email in a poll (it uses .first()), use from.text for the sender, and add JSON parse error handling and a spam filter.

Try it yourself with n8n

Build an n8n workflow that watches [support inbox] and, for each new email, classifies the topic, assigns a priority of low, medium or high, and writes a two-line summary. Have it draft a reply using [link to FAQ or policies] and post the summary and draft to [Slack channel] for a person to review. It must never send the reply itself; you're done when a test email produces a Slack post I can approve.

1 more account of this

Read the original ↗

Source: community.n8n.io · Undated

Five of these in your inbox every morning

The best things people got an AI agent to do, each with the prompt to try it.

More like this