New generation media spaces let group members see each other and share information. However, they are separate from the real world; participants cannot see beyond the video, and they cannot engage with people not attending to the computer. To solve this problem, we use a robot as a physical surrogate for a media space group, which allows this distance-separated group to extend their interactions into the real world. Through video, all media space group members see a first-person view of what the robot sees. All have opportunity to control it: where it walks, where it looks, and even the sound it makes. The robot becomes a physical tele-embodiment of the group, representing it for people who may not physically be part of the group but are collocated with the robot.