Automated Issue Creation without User Confirmation in Rovo and Virtual Service Agent

[João Cunha]

Hi,

In our company we are currently in the testing and adoption phase of Atlassian Rovo and JSM Virtual Service Agent. We have successfully connected our agents to our Confluence knowledge base, but we are facing challenges regarding the automation of ticket creation.

Goal: We want to ensure that every interaction results in a Jira Service Management ticket (Project: ROVO) for audit and history purposes, without requiring additional user confirmation (eliminating the "Raise Request" button or manual confirmation steps).

Scenarios:

1. Successful Resolution: User states a problem -> Agent provides a solution from Confluence -> User confirms it's resolved -> A ticket is automatically created and closed/done without further user intervention.
2. Escalation/Unresolved: User states a problem -> Agent provides solutions that don't work (or no solution) -> User requests a ticket -> Ticket is created automatically without the user having to "confirm" or "edit" the request.

Current Issues:

• Rovo Agents: Even when instructed to create a ticket, the agent consistently displays the "Raise Request" button/confirmation card, which adds friction to the user experience.
• Virtual Service Agent (VSA): * If using the default AI answer, once the user clicks "Yes" to "Did this resolve your issue?", the agent jumps straight to the CSAT survey without creating a ticket.
  • If we use Intents (e.g., "help", "problem"), the agent stops searching Confluence automatically and goes straight to our manual flow questions.

Questions:

1. Is there a way to suppress the "Raise Request" confirmation in Rovo for background actions?
2. How can we configure VSA to trigger an "Issue Creation" action automatically after a successful AI-generated answer?
3. Can we integrate a Knowledge Base search node within a manual Intent Flow so we don't lose the AI's ability to search Confluence?

Looking forward to your guidance.

Best regards!

[Dr Valeri Colon _Connect Centric_]

Hi @João Cunha Rovo and the Virtual Service Agent can’t create JSM tickets silently today. Both require an explicit confirmation step—the “Raise Request” card in Rovo and an Intent-triggered action in VSA. These safeguards are part of JSM’s design to avoid unwanted ticket creation. You can’t suppress the card, auto-create after a resolved answer, or embed KB search inside a manual intent flow. The only workaround is Automation or a Forge app that creates tickets outside the chat flow. The pattern is:

1. User chats with Rovo/VSA → the agent posts a structured message (labels, keywords, or a hidden marker) into the ticket or conversation.
2. Jira Automation watches for that marker (e.g., a comment containing “AUTO_CREATE”) or for a specific transition like “Solution confirmed.”
3. Automation rule creates the JSM ticket immediately, fills fields from context, and optionally closes it.

Rovo supplies the text; Automation performs the actual creation.

[João Cunha]

Hi Dr Valeri Colon _Connect Centric,

Thank you for the clear explanation regarding the "silent" ticket creation limitations. I understand that the confirmation step is a safeguard.

Regarding your workaround suggestion: "User chats with Rovo/VSA → the agent posts a structured message (labels, keywords, or a hidden marker) into the ticket or conversation."

I have a follow-up question: How exactly can Jira Automation "listen" or "catch" a specific keyword or hidden marker posted inside a Rovo/VSA chat session?

Since the conversation hasn't become a ticket yet (in the Rovo scenario), I couldn't find a trigger in Jira Automation that monitors live chat messages for specific strings.

Is there a specific "Issue Comment" or "Chat Message" trigger that supports Rovo/VSA logs?

Or are you suggesting that the agent should post this to an external endpoint via a Webhook (which I currently don't see as an available Skill in my Rovo Agent)?

Any documentation or specific trigger names for this "keyword monitoring" would be extremely helpful.

Best regards,
João Cunha