we could hold our own database, similarly as we create our own code replacements before the event we could input images with associated names
That would save me so much time at events, even if only some of them got done, and suggested names for images not fully identified.
Or we could select the face with the crop tool and then identify the face.
I believe that the two apps you mentioned are "trained" by the user. There is no centralized database that is consulted by the feature.
I would pay more money for that add-on