So far as I am aware that is not possible. You have to pre-train everything that MOVI understands. Maybe if you were able to train all names of all possible users of an application you could do something like that.
In my experience, MOVI works best with a traditional call-sign and list of commands type of application. Ie. you say something like “OKAY MOVEE”, MOVI goes Beep-Beep, then you tell it what you want it to do based on a predetermined menu of options.