This is a new series of unscripted, unplanned videos going through some testing and exploration of Copilot in Power BI. There is no editing or salesmanship here; it's just straight-up what I see and find. I'm not trying to entertain anyone or grow a following.
This series will contain my observations and opinions as I explore Copilot in Power BI each month. I'll share things I think are useful to know if you use Copilot or make reports or models that people will use -with- Copilot. In this series I'm using a representative model and report for Order-to-Cash data with multiple fact tables that is similar to ones I see in organizations.
00:00 Introduction
01:37 Troubleshooting strange Copilot outputs
05:00 Exploring data from Copilot outputs
06:21 Users asking for sensitive data
Copilot and its capabilities are evolving over time. I feel it is important to be skeptical and constructively critical of this technology since it is new and can be unreliable. I think there are valid use-cases and in general find it has improved since I last used it in September. But I'm concerned that there are still a lot of unrealistic expectations about what it can and should do, and what a model author needs to do to get Copilot to "work" the way users want.