Support Ticket Classifier Tutorial
Build an AI-powered support ticket classifier with ActionFlows. A step-by-step tutorial that routes and prioritizes incoming tickets automatically.
AI-Powered Support Ticket Classifier
Learn ActionFlows through real-world examples. Each tutorial takes 15-20 minutes and builds a production-ready workflow.
What you'll build: A system that automatically categorizes incoming support emails and routes them to the right team.
Time: 15 minutes
Skills you'll learn: Start node, AI classification, conditional routing, Slack notifications
The Problem
Your support team receives 100+ emails daily. Manually sorting them into categories (Bug, Feature Request, Billing, General) wastes 2 hours per day.
The Solution
A workflow that reads each email, uses AI to determine the category, and sends it to the appropriate Slack channel.
Step 1: Create the flow (2 minutes)
Create a new flow called Support Email Classifier.
Add a Start node (manual). There is no Email start-node type.
For this tutorial, run the flow with sample email text in the run form. In production, attach a flow webhook from your helpdesk or mailbox so each new ticket POSTs into the same graph.
Leave the Generate Text Prompt empty (or add a mapped field) so Run flow asks for the email body as {nodeId}_prompt. Studio autosaves. Open Settings and click Set to active.
Step 2: Add AI Classification (3 minutes)
Add a Generate Text node. Connect it to your Start node.
Configure the prompt:
You are a support email classifier. Read this email and determine which category it belongs to.
Email Subject: Use the email subject line
Email Body: Use the email body content
Categories:
- BUG: Technical issues or errors
- FEATURE: Feature requests or suggestions
- BILLING: Payment or subscription questions
- GENERAL: Other inquiries
Respond ONLY with the category name (BUG, FEATURE, BILLING, or GENERAL).
Do not include explanation. Just the category.This node will output just the category name.
Step 3: Create Conditional Routing (5 minutes)
Add a Switch node (also called Conditional). Connect the AI output to it.
In the Switch configuration:
Create four cases:
Case 1: If output equals BUG
Case 2: If output equals FEATURE
Case 3: If output equals BILLING
Case 4: Default (anything else)
Step 4: Add Team Notifications (5 minutes)
For each case, add a Slack node.
For BUG case:
- Channel:
#bugs - Message:
Bug Report from Support
From: Use sender email
Subject: Use email subject
Category: BUG
Message: Use email body previewFor FEATURE case:
- Channel:
#feature-requests - Message:
Feature Request
From: Use sender email
Subject: Use email subject
Category: FEATURE
Details: Use email body previewFor BILLING case:
- Channel:
#billing - Message:
Billing Inquiry
From: Use sender email
Subject: Use email subject
Category: BILLING
Details: Use email body previewFor GENERAL case:
- Channel:
#general-support - Message:
General Support Email
From: Use sender email
Subject: Use email subject
Message: Use email body preview// IMG: Complete flow with all four Slack branches
Step 5: Add Email Reply (Optional but Recommended) (3 minutes)
Add an Email node to each branch. Configure it to send an automatic acknowledgment:
Thank you for contacting support!
We have received your message and routed it to the appropriate team.
You will hear back within 24 hours.
Best regards,
Support TeamStep 6: Test the Workflow (2 minutes)
Click Test.
Provide sample email data:
- Subject: "Login button not working on mobile"
- Body: "I cannot log in using the mobile app. Error says 'Invalid credentials.'"
Run the test. The workflow should:
- Read the email
- Classify it as BUG
- Send to
#bugschannel - Send acknowledgment email
Check your Slack and email to confirm.
Step 7: Activate and attach inbound HTTP
Studio already autosaves. Open Settings and click Set to active.
This graph still starts from Start. It does not watch a mailbox. For production tickets, create a flow webhook and POST each email payload from your helpdesk, or keep using Run flow with sample text.
What's Happening
Start (or flow webhook POST)
↓
Generate Text classifies
↓
Switch routes
↓
Slack (named connection)↓ Auto-reply email sent
Your support team now sees tickets in organized channels, not a chaotic inbox.
---
### Next Level: Add More Features
**Count tickets by category:** Use a Database node to log each classification for analytics.
**Escalate urgent issues:** Add an If node to check for keywords like "urgent" or "critical" and notify a manager.
**Learn and improve:** Export classification logs and refine your prompt based on misclassifications.