When the AI Agent can't help a customer, it gracefully escalates to your human support team. This ensures customers always get the help they need.

When AI Escalates

The AI will offer escalation when:

Can't Find an Answer

The knowledge base doesn't contain information to answer the question.

Example:

Customer: "What's your refund policy for annual subscriptions?" AI: "I don't have information about annual subscription refunds in our documentation. Would you like me to connect you with our support team?"

Customer Requests Human Help

The customer explicitly asks to talk to a person.

Example:

Customer: "I need to speak with someone" AI: "Of course! I'll help you connect with our support team."

Question Requires Account Access

The question needs customer-specific information the AI doesn't have.

Example:

Customer: "Why was I charged twice this month?" AI: "I don't have access to account-specific billing information. Let me connect you with our support team who can look into this."

Complex or Sensitive Issues

Some topics are better handled by humans.

Example:

Customer: "I want to cancel my account and get a full refund" AI: "I understand. For account cancellation and refund requests, our support team can help you directly."

The Escalation Flow

Step 1: AI Recognizes Escalation Need

The AI determines it can't adequately help and offers to escalate.

Step 2: Customer Confirms

The customer agrees to create a support ticket.

Step 3: Escalation Form

The customer provides:

Their email address (if not already known)
Subject line (AI may suggest based on conversation)
Any additional details

Step 4: Ticket Created

A support ticket is created with:

Customer's information
Subject and description
Full conversation transcript from AI chat
AI-suggested tags (if applicable)

Step 5: Team Notified

Your support team receives the ticket through normal channels:

Appears in ticket inbox
Notifications sent (Slack, email as configured)
Auto-routing applies if tags trigger it

What Gets Included in Escalated Tickets

Conversation Transcript

The complete AI conversation is attached to the ticket, so your team sees:

What the customer originally asked
What the AI responded
Where the conversation went
Why escalation was needed

This context helps your team respond faster and more accurately.

Suggested Tags

Based on the conversation, the AI may suggest relevant tags:

"Billing" for payment questions
"Technical" for product issues
"Account" for login/access problems

Tags can trigger auto-routing to the appropriate team.

Benefits of AI Escalation

For Customers

No repeat information - Conversation context is preserved
Smooth handoff - Clear transition from AI to human
Always an option - Never stuck with only AI
Quick start - Ticket created with context already captured

For Your Team

Pre-qualified issues - AI handled simple questions first
Full context - See what was already discussed
Relevant routing - Tags help tickets reach the right team
Reduced volume - Only true escalations come through

Monitoring Escalations

Track Escalation Rate

Monitor what percentage of AI conversations escalate:

Low (< 20%): AI is handling most inquiries well
Medium (20-40%): Normal, some topics need human attention
High (> 40%): May indicate KB gaps or AI limitations

Analyze Escalation Reasons

Review escalated tickets to identify patterns:

Pattern	Action
Common question without KB answer	Write an article
Unclear existing article	Improve the article
Account-specific questions	Expected (can't improve)
Frustrated customers	Review AI tone/responses

Reduce Unnecessary Escalations

Identify top escalation topics from ticket analysis
Write or improve articles covering those topics
Monitor if escalations decrease for those topics

Escalation Settings

Escalation Prompt

The message customers see when AI offers escalation. Customize tone and wording in settings if available.

Tag Suggestions

Configure which tags the AI can suggest during escalation. This improves routing accuracy.

Escalation Triggers

The AI automatically determines when to offer escalation. You can't configure specific triggers, but you can influence behavior by:

Adding more comprehensive KB content
Ensuring articles cover edge cases
Writing clear, searchable article titles

Best Practices

Make Escalation Easy

Don't make customers work to reach humans:

Keep the escalation form simple
Don't require excessive information
Acknowledge their time with AI

Respond to Escalations Quickly

Customers who escalated may already be frustrated:

Prioritize escalated tickets
Reference the AI conversation in your response
Show you've read the context

Learn from Escalations

Every escalation is an opportunity:

Identify missing KB content
Find confusing documentation
Understand customer pain points