Ticket escalation

Ticket escalation in Eddy AI Chatbot lets users raise support tickets directly from the chatbot interface when the chatbot cannot assist them. When you connect Zendesk or Freshdesk, the chatbot automatically detects situations where a user needs human support and prompts them to submit a ticket — pre-filled with the chat transcript, an AI-generated subject line, and an AI-generated issue description that the user can edit before submitting.

When to use ticket escalation

• Use this when you want to give users a fallback option within the chatbot, so they do not need to leave the chat to contact support.
• Use this when users frequently encounter questions the chatbot cannot answer, and you want those interactions to generate actionable support tickets automatically.
• Use this to capture structured ticket data from chatbot conversations — including the full transcript — so agents have complete context when they pick up the ticket.

Before you begin

• You must have created an Eddy AI Chatbot with at least one source connected.
• Have your Zendesk or Freshdesk API credentials ready:
  • Zendesk: API Token, Administrator Email ID, and Subdomain URL.
  • Freshdesk: API Token and Subdomain URL.
• Ticket escalation currently supports Zendesk and Freshdesk only.

How to configure ticket escalation

1. Navigate to AI Chatbot () in the left navigation bar and click Customize.
2. Navigate to the Ticket escalation () tab.

3. Click Connect next to your preferred ticketing platform:
   Zendesk or Freshdesk.

4. Enter your connection details:

   For Zendesk:

   • API Token — your ticketing platform's API token for authentication.
   • Subdomain URL — your ticketing platform subdomain.
   • Administrator Email ID (Zendesk only) — the admin email address for authentication.

   

   For Freshdesk:

   • API Token — your Freshdesk API token for authentication.
   • Subdomain URL — your Freshdesk subdomain.

5. Once connected, the chatbot automatically prompts users to raise a ticket when any of the following occur:

   • A user marks a chatbot response as unhelpful.
   • The chatbot cannot find an answer.
   • The chatbot detects negative sentiment or frustration in the conversation.
   • A user explicitly says something like "raise a ticket" or "I need support."

6. When a ticket is triggered, the ticket form is pre-filled with:

   • The full chat transcript.
   • An AI-generated subject line.
   • An AI-generated issue description (editable by the user before submission).

7. Once the user submits the ticket, the ticket ID and details are shared back in the chat.

Suggest a ticketing platform

If you use a platform other than Zendesk or Freshdesk, you can submit a suggestion to have it added:

1. Click Suggest more ticketing platforms.
2. Select from the available options:
   • Jira
   • Intercom
   • GitHub
   • HubSpot
   • Others (enter the platform name and the purpose of integration)
3. Submit your suggestion to help us add more integrations.

Best practices

• Connect ticket escalation before going live — setting up ticket escalation at launch ensures users always have a path to human support from day one.
• Test the escalation flow in the Playground — ask the chatbot a question it cannot answer and verify that the ticket escalation prompt appears correctly.
• Use Zendesk or Freshdesk as both a source and an escalation target — if you connect Zendesk or Freshdesk as a training source (via the Source tab), the chatbot learns from past tickets. Connecting it for escalation routes new unresolved queries back to your support team, closing the loop.
• Review escalated tickets to improve the chatbot — analyse escalated tickets regularly to identify common questions the chatbot could not answer. Add that content as a Text, FAQ, or File source to reduce future escalations.