This is right! Identity Segmentation is something we’ve been working on for some time and continue to work on. What’s being described in this thread (Segmentation based on source) should be a use case fairly easily covered so long as there is an identity attribute that is also connected to that source. For example, an employee type identity attribute “Contractor” for NERM and “Employee” for Workday.
As you can imagine, there’s many, many places identities/users are exposed compared to other objects, so identity segmentation takes a bit longer for us to build.