Allow Intelligent Triage intents to use all ticket interactions

Add an option to let Intelligent Triage intents evaluate all relevant ticket interactions, not only public messages from requesters, but also agent replies.

Currently, intent detection is based only on public messages written by customers. These messages are often incomplete, unclear, or describe symptoms rather than the underlying issue. Agents frequently clarify the actual problem in their replies after investigation.

While agents can manually change the intent, this is a limited workaround:

• It removes much of the time-saving benefit of automated intent classification.
• It introduces agent bias and inconsistency.
• It increases the risk of errors.
• It requires agents to understand and navigate the full intent taxonomy, which becomes harder as intents are added, updated, or deactivated.
• Deactivated intents still appear in the dropdown, making manual selection slower and more error-prone.

As a result, intent classification can remain inaccurate even after the issue is fully understood, because the root cause is often identified by the agent rather than clearly stated by the customer. This leads to:

• Routing and automation based on incomplete or misleading information.
• Reporting and trend analysis reflecting customer-described outcomes instead of actual causes.
• Reduced value of Intelligent Triage for complex or technical issues, where accurate categorization depends on agent investigation.

Providing an option to include agent replies in intent evaluation would allow intents to reflect the true nature of the ticket, not just the initial customer description. This would improve accuracy while preserving the benefits of automation.

There is currently no effective workaround, as intent logic cannot be configured to consider agent-written content.