AI agent tickets are tickets for conversations that were handled entirely by a messaging or voiceAI agentor by adefault messaging response, meaning they haven’t yet been or never were escalated to a human agent. Agents and admins can view AI agent tickets in the Agent Workspace.

What's my plan?
All Suites Team, Growth, Professional, Enterprise, or Enterprise Plus
Support Team, Professional, or Enterprise

AI agent tickets are tickets for conversations that were handled entirely by a messaging or voice AI agent or by a default messaging response, meaning they haven’t yet been or never were escalated to a human agent. Agents and admins can view AI agent tickets in the Agent Workspace.

AI agent tickets help you:

  • Identify customer requests that are ongoing or have been solved without human intervention. Search for or create views for AI agent tickets to quickly discover which types of issues are being solved without human agent involvement.
  • Provide context for customer interactions. AI agent tickets appear in the interaction history for customers who have had previous interactions with the AI agent. This gives agents who are handling a ticket more context on the engagements that this customer has had with your business.
  • Audit conversations to analyze AI agent responses. Read the full conversations between your AI agent and customers to see how your AI agent is responding and identify areas for improvement.
  • Gain better visibility into your automated resolutions. Examine your solved and closed AI agent tickets to confirm that interactions counted as automated resolutions are being resolved appropriately.

 

Viewing AI agent conversations as tickets in the Agent Workspace (4:17)

This article contains the following topics:

  • About AI agent tickets
  • Turning the AI agent tickets feature off (AI agents - Advanced only)
  • Identifying an AI agent ticket
  • Understanding functionality limitations for AI agent tickets

Related article:

  • Finding AI agent tickets in your account
  • Interacting with AI agent tickets

About AI agent tickets

AI agent tickets are created for AI agent functionality, legacy AI agent functionality, third-party bots, and default messaging responses. When a customer begins a conversation with a messaging or voice AI agent (or just types in the Web Widget), an AI agent ticket is created.

AI agent tickets cannot be edited like regular tickets. For admins and agents with permission, an AI agent ticket's tags, custom fields, and followers are editable, and internal notes can be added. For all other users, AI agent tickets are read-only. If necessary, an admin or agent with permission can take over the ticket at any point during the AI agent conversation.

As long as a conversation is not escalated to a human agent, it is considered an AI agent ticket. When an AI agent ticket is escalated to a human agent, the ticket is no longer considered an AI agent ticket and becomes a regular, editable ticket.

AI agent tickets don’t disrupt existing workflows based on business rules, including automations, ticket triggers, and more. These types of workflows continue to function only on regular tickets and cannot act on AI agent tickets. This means that these workflows take effect as expected when an AI agent ticket is escalated to a human agent. If an AI agent ticket is escalated, its status is set back from Open to New, allowing it to be processed as any other newly submitted ticket would be.

For AI agent tickets, tags based on users and organizations are populated at ticket creation, but any ticket triggers or automations based on these tags don’t fire until a ticket is escalated to a human agent.

AI agent tickets count toward your account’s data storage limits. If necessary, you can create a ticket deletion schedule to target AI agent tickets specifically.

About AI agent tickets for messaging AI agents

In messaging conversations, the first customer message in the conversation triggers the creation of the AI agent ticket. This means any initial message sent by the AI agent or basic message response (for example, a welcome message) is not included. However, if the conversation is escalated to a human agent, then the conversation history is backfilled and any initial messages are inserted in the ticket upon escalation.

While an AI agent conversation is ongoing, the AI agent ticket is in the Open status. The conversation is no longer considered ongoing when:

  • 4 days have passed since the last response, or
  • The AI agent releases control, which marks the conversation as Solved, or
  • (For third-party bots only) The third-party bot logic deems the conversation resolved.

At that point, an AI agent ticket moves into the Solved status, and then into the Closed status based on the Close ticket 4 days after status is set to solved automation (which is the only automation that works on AI agent tickets).

AI agent tickets can be deleted only when they’re in the Solved or Closed status. This prevents agents or admins from accidentally deleting the ticket for an ongoing AI agent conversation.

Deleting an AI agent ticket doesn’t delete the underlying Sunshine Conversations data associated with that conversation. In a future release, Zendesk plans to introduce the ability to remove these conversations from Sunshine Conversations too.

Messages sent by a suspended user don’t create an AI agent ticket. If you need to delete these messages for General Data Protection Regulation (GDPR) or other compliance reasons, create a deletion schedule for bot-only conversations.

How metadata is added to AI agent tickets

By default, conversations have no metadata. However, you can use either the Web Widget API or the Sunshine Conversations API (for third-party bots) to apply metadata to a conversation.

Metadata can be set on an AI agent ticket during any of the following events:

  • When the conversation is created. If metadata is present when a conversation is created, that metadata is mapped to the AI agent ticket at creation time.
  • When the AI agent escalates the conversation to a human agent. If metadata is present when a passControl action occurs, the metadata is mapped to the AI agent ticket at that point. This is also when the ticket transitions from read-only to fully editable.
  • When the session ends or the ticket is solved. If metadata is present when a releaseControl action occurs, all metadata is automatically backfilled to the ticket. There’s no need to pass it explicitly using the API.

About AI agent tickets for voice AI agents

In voice conversations, the AI agent ticket is created when the call connects.

After the AI agent has finished the conversation, the AI agent ticket is converted into a regular, editable ticket to allow for any needed follow-up. The ticket isn’t automatically resolved, and must be resolved or closed based on your ticket triggers or automations as normal. This means that, unlike AI agent tickets for messaging tickets, which can remain AI agent tickets as long as no human agent ever needs to be involved, AI agent tickets for voice will always become a regular ticket.

Additionally, there is no pass control or release control for voice tickets. Every AI agent ticket is converted to a regular ticket after the AI agent finishes the conversation. Resolution types are not populated.

Turning the AI agent tickets feature off (AI agents - Advanced only)

Note: In most accounts, the AI agent tickets feature is turned on in your account and cannot be turned off, meaning the Turn on AI agent conversations as tickets setting does not appear.

If you use AI agents - Advanced and have previously configured a custom ticket creation flow, you can turn the AI agent tickets feature off to prevent duplicate tickets from being created for conversations between your AI agents and your customers.

To turn the AI agent tickets feature on or off

  1. In Admin Center, click Channels in the sidebar, then select Messaging and social > Messaging.
  2. Click Manage settings.
  3. Click the AI agent conversations as tickets in Agent Workspace section to expand it.
  4. Select or deselect Turn on AI agent conversations as tickets.

  5. Click Save.

Identifying an AI agent ticket

To help you distinguish between AI agent tickets and regular tickets, AI agent tickets have a number of visual identifiers:

  1. A bot icon appears in the ticket tab.
  2. An “AI agent” badge appears in the ticket header (and the filter is missing).
  3. All the fields in the ticket properties panel are grayed out.
  4. The composer and macros menu are hidden.
  5. A message in the ticket footer notes that this is an AI agent ticket.

Additionally, AI agent tickets include helpful indicators when an automated resolution is consumed:

  • The Resolution type field lists Automated as the value.
  • In the events view, a system entry notes the date and time the automated resolution occurred.
  • The Resolution type field can be used when creating custom reports.

Note: Deleting an AI agent ticket with a Resolution type value of Automated doesn’t undo the consumption of an automated resolution.

To help agents and admins identify AI agent tickets in views and search results in Support, the Assignee column includes a bot icon for AI agent tickets.

In Support, the profile associated with the AI agent assignee has the Contributor role.

For more information, see Finding AI agent tickets in your account.

Understanding functionality limitations for AI agent tickets

AI agent tickets have different limitations and considerations than regular tickets. The following sections provide more details:

  • Functionality limitations
  • Sunshine Conversations considerations

Functionality limitations

The following functionality does not currently work on AI agent tickets until they are escalated to a human agent (at which point they become regular, editable tickets):

  • Deleting tickets for ongoing AI agent conversations
  • Modifying ticket properties
  • Adding public comments or internal notes
  • Applying macros
  • Ticket triggers and automations (except for the Close ticket 4 days after status is set to solved automation)
    • Messaging triggers do work on AI agent tickets
  • SLAs
  • Omnichannel routing
  • Skills calculation
  • Ticket assignment/reassignment
  • Bulk editing
  • Ticket sharing
  • Custom ticket statuses
  • Problem/incident ticket linking
  • Exporting ticket data
  • Ticket merging
  • Auto assist (including actions)
  • Automatic redaction

    • Manual redaction and redaction suggestionsdo work on AI agent tickets. Note that redaction suggestions work in ticket comments only, not in PDF attachments.

  • Workflows that rely on the “created_at” event
  • Domain-level events in integrations
  • Zendesk Apps framework (ZAF) apps

Additionally, AI agent tickets don’t appear in the following places:

  • Agent Home
  • Explore reports (except for the Resolution type field)
  • The outputs of APIs that list tickets (such as the endpoints listed here)
    Note: By default, these exclude AI agent tickets, but you can retrieve them specifically by adding the ?support_type_scope=ai_agent query parameter (supported on /api/v2/tickets and /api/v2/incremental/tickets).

Sunshine Conversations considerations

The following Sunshine Conversations considerations apply to AI agent tickets:

  • Pre-existing conversations are not backfilled as AI agent tickets.
  • Each new zd:agentWorkspace conversation generates a regular ticket.
  • Each new non-zd:agentWorkspace conversation generates an AI agent ticket.
  • User data is synced either when an AI agent ticket is marked as Solved or when the ticket is escalated to an agent.
  • The releaseControl action marks the associated AI agent ticket as Solved.
Powered by Zendesk