Julius skips the DAX and the dataset
Inside a Power BI organization, a lot of analysis still happens in Excel: a file arrives from a vendor or a department, and someone needs an answer before anyone could model it. Julius fits that gap. Upload the workbook or connect a database, type the question, and Julius produces and executes Python, R, or SQL, then explains the chart it produced. There is no data model to design, no DAX measure to write, and no report to publish before the answer appears.
Nothing Julius produces becomes part of the governed reporting estate, though. It sits outside Power BI's workspaces and Microsoft-managed security, and nothing it produces refreshes on a schedule. An insight found in Julius has to be rebuilt in Power BI if it should reach executives through the usual channel.