There is already a quite long roadmap for the app, that goes from UI improvements, like a method for classifying the geometry organized into collections, to some improvements on the optimization of the location for the panos. Also some computer vision algorithms that help and fasten the reference process, and a long ...
Nowadays the software is limited to work with models where all panos belongs to the same ground, so by now it's no possible to create a single model with multiple levels. But that of course will be implemented at some point.
There are some inner tools that are only available in mac version, like the vertical align tool for panoramas, or the possibility to render a tif depthmap image.
It's also not possible to perform undo / redo actions, even it's in the menu, those are just for use in the forms.