Problem
Support teams manually review every inbound ticket to decide routing, a process that breaks under volume, misses P1 escalations during off-hours, and gives high-churn customers the same response time as routine questions.
Solution
An n8n workflow that acts as a first-responder for every inbound support ticket. Classifying across 4 dimensions (priority, category, churn risk, VIP status) before any human sees it. CRITICAL and URGENT tickets trigger Slack escalation and agent email with full context. STANDARD tickets get an instant Claude-drafted response. All decisions logged to Sheets for SLA tracking. Zero manual triage steps.
Demo
Results
- 21 nodes across email and webhook input channels
- 4-dimension AI classification. Priority, category, churn risk, VIP status
- CRITICAL escalations auto-routed to Slack + agent email
- Claude-drafted auto-responses for P3 standard tickets
- Full SLA tracking via Google Sheets audit log
Webhook and email triggers converge at a single classification node where Claude evaluates ticket text against a structured scoring rubric. Claude returns a JSON object with priority, category, churn_risk (boolean), and vip_status (boolean). Downstream switch nodes branch on the combined scoring, not just priority alone, to handle nuanced routing (e.g., a STANDARD ticket from a high-churn VIP customer gets a Slack flag even though it's not CRITICAL). Gmail and Slack nodes handle all outbound communication. A final Sheets append node logs every decision with the full classification and estimated SLA window.
Tech Stack
- n8n
- Claude API
- Slack API
- Gmail / SMTP
- Google Sheets API