Virtual agent vs Rovo agent - Which to use?

[Rhys Duindam]

Heya, I have a question about which agent is best to use? Currently the support page gets cluttered quickly with the pop-up, but the virtual agent seems to have far inferior results to the Rovo agent. But Flows and data gathering is possible in virtual agents. All in all I am just confused which I am supposed to use. What are other people doing?

Cheers,

Rhys

[Fabio Racobaldo _Catworkx_]

Hi @Rhys Duindam and welcome,

Rovo and virtual agents are differents and could be used for different purposes.

Atlassian does not provide a single “best” choice — each agent was designed for different use‑cases.

Rovo agents are AI teammates that specialize in tasks like editing content, analyzing input, drafting responses, and using connected knowledge sources.
They can search across Jira, Confluence, and third‑party integrations for deep context.

They are not designed for structured workflows or form‑based user journeys — only for conversational help and intelligent knowledge tasks.

The Virtual Agent is part of Jira Service Management’s customer intake automation.
It focuses on guided conversations, conditional logic, “flows,” decision trees, and structured data capture — things Rovo does not support.

Virtual Agent does not have Rovo’s depth of AI knowledge; instead it’s optimized for repeatable workflows.

Generally, my suggestion is to use Virtual Agent for customer-facing flows and Rovo Agent internally.

Hope this helps,

Fabio

[Rhys Duindam]

Thanks @Fabio Racobaldo _Catworkx_ thats an appreciated advice! I have immediately leant more to virtual agent. It looks much cleaner that way. Shame I can't define anything behind the rovo integration e.g. Identity. But its better for data gathering and improving the service. Thanks again!

[burakgeyik]

Hello @Rhys Duindam, all,

I just came accross to this great subject - this is a very hot topic and this is where I belive the limits of AI Agents are tested in our field.

We are currently working on a beta application with one international customer. Our application is a kind of "combination" of Rovo + Virtual agents:

• You decide from which sources your would like to ingest information from
• You decide which work items your AI Agent should work on/collaborate with
• You decide how your AI Agent acts on work items
• Or you let your AI Agent decide how to act. 

To be specific, this is where you define "what" and "how":

In the above example, you instruct your agent:

• To work only on the work items on "Customer Support Portal 14" 
• To work only on the work items with the "Waiting for Support" status
• To work only on the work items from the customers with "Elite" Support package.

If you uncheck all the check boxes, you AI Agent works only internally, helps your service desk agents by preparing all the data and any follow up questions if needed; and there wont be any visibility to the external customers. 

You are welcome to see more about the Technical Support AI Agent here:

Technical Support AI Agent by SolveroX

And the documentation to get more insight about its capabilities: Technical Support AI Agent - Documentation 

If you are interested in joining the beta - just let us know. 

One side note: If you are interested in the most critical insights into JSM service quality, this is our first application - now available free for an extended period. It enables one-click reporting on the most important JSM metrics, it is worth to check:
https://marketplace.atlassian.com/apps/2099091683/metrics-and-reports-for-jira-service-management-jsm 

Kind regards.