On a spot by spot basis, you could always use the identify tool to get your pixel value for individual pixels, which would be the height of the object at that point. I tend to call a DSM minus a DEM a Digital Height Model(DHM) for it is the height of the objects scanned above ground(bare earth).
Differentiating between trees and other high features can be challenging, though research has been done for classifying different feature types from LiDAR. If all they need are to be able to get the heights on a case by case basis, then the DHM and the identify tool to get the pixel value would be all they need.