Using AI Ticket Classification in Helpdesk

Helpdesk’s AI automatically classifies every incoming ticket — detecting spam, sentiment, intent, language, and suggested priority. In this tutorial, you’ll learn how to use these predictions and improve accuracy with custom intents and agent corrections.

Prerequisites

• Agent or admin role in Helpdesk
• Incoming tickets (AI classification runs automatically on new tickets)

Step 1: Review AI predictions on a ticket

1. Open any ticket
2. Look for the AI classification section in the ticket detail panel
3. You’ll see:
   • Spam score — 0-100 likelihood of spam
   • Sentiment — positive, neutral, negative, or frustrated
   • Intent — what the customer wants (e.g. “billing inquiry”)
   • Language — detected language
   • Suggested priority — AI-recommended priority
   • Confidence — how certain the AI is about each prediction

Step 2: Create custom intent categories

Tailor the AI to your business vocabulary:

1. Navigate to Settings > Intent Categories
2. Click New Intent (or start with a Starter Pack for your industry)
3. Enter:
   • Name: “Account Cancellation”
   • Slug: account_cancellation
   • Description: “Customer wants to cancel their account or subscription”
   • Examples (3-10 required):
     • “I want to cancel my subscription”
     • “Please close my account”
     • “How do I stop my membership?”
     • “Cancel my plan effective immediately”
     • “I’d like to terminate my service”
4. Save

The AI will now recognize this intent on future tickets.

Step 3: Correct a prediction

When the AI gets something wrong:

1. Open a ticket with an incorrect intent or language
2. Click the field you want to correct
3. Select the correct value from the dropdown
4. The correction is saved immediately

Both the original AI value and your correction are preserved. Corrections appear in the ticket activity feed.

Step 4: Build automations with AI fields

Use classification results to power automations:

Example: Auto-close spam

1. Create a trigger with condition: Spam score → greater than → 90
2. Add action: Set status → closed

Example: Escalate negative sentiment

1. Create a trigger with condition: Sentiment → is → negative
2. Add action: Set group → your escalation team

Example: Route by language

1. Create a trigger with condition: Language → is → es
2. Add action: Set group → your Spanish-speaking team

Tips

• Provide diverse examples for custom intents — the more varied your examples, the better the AI recognizes the intent
• Monitor corrections — if agents frequently correct the same intent, your examples may need updating
• Start with starter packs — they give you a solid baseline that you can customize
• AI runs asynchronously — classification happens in the background. If the AI service is temporarily unavailable, tickets are retried automatically

What’s next

• AI Ticket Classification reference — full documentation
• Intent Categories — manage intents in detail
• Agent Corrections — how corrections work