Accuracy of ROVO and limitation on number of records in report generation

[Raghuveer Madhu]

Hi Team,
I am currently using ROVO and had a couple of questions:

What is the present accuracy level of ROVO when generating reports?

I have observed a limitation in the number of records it pulls while generating a report. Is there a way to improve or configure this to handle larger datasets?

Any guidance or best practices from the community would be much appreciated.

I really appreciate any help you can provide.

[Dr Valeri Colon _Connect Centric_]

@Raghuveer Madhu Rovo doesn’t have a fixed “accuracy rate”—it depends entirely on the data you give it and how well the agent instructions constrain the output. For report generation, Rovo is capped by result-size limits in Search and Jira queries, so it won’t return very large datasets. There’s no config to lift that limit yet; the workaround is chunking queries or narrowing the scope.

[Rebekka Heilmann _viadee_]

Hi @Raghuveer Madhu 

Report as in Deep Research? Or general Search/Chat?

Atlassian has not revealed any specifics on parameters on the underlying models. There currently is no way for us users to change any of the parameters (no. of sources, creativity etc.). Rovo is fairly non-deterministic and uses multiple LLMs in the background so qualitiy in results may vary significantly.

They have however published information on the algorithm used and how relevance of results is measured: Unraveling Rovo search relevance - Work Life by Atlassian