I can think of a few useful applications that have nothing to do with gesture recognition; the z-depth would allow you to cut the image of the user away from the background, which could be useful for videoconferencing if you want to chat with someone but don't want them to see your surroundings or the other people in the room with you.
Close-field Kinect on a laptop would probably allow you to do very accurate head / eye tracking, so you can do things like the '3d window' effect demonstrated in this video by Jonny Lee.
3D object scanning. Microsoft already have a version of this for the Xbox with the Kinect Fun Labs, but it's rather limited in resolution. A close-field sensor would provide better resolution.
It's possible you could use facial recognition for security. Unlike most webcam-based facial recognition systems, this couldn't be fooled by a paper mask.