One of the ideas of ubiquitous computing is that computing resources should be embedded ubiquitously in the environment, making them available to any nearby users. Some researchers have applied this to interaction, and tried to embed an abundance of interactive devices, such as touch screens, in rooms and whole buildings. The opposite concept is that of a single personal mobile device, which users carry at all times and use for all interactions. Because both concepts have different strengths, we explore building interfaces for federations of personal mobile and stationary embedded devices, exploiting the capabilities of both rather than forcing users to choose between either. We have developed an infrastructure that coordinates multiple devices for that purpose: groups of devices work together to render a user interface. As one of the main challenges for such federated user interfaces we have identified their authoring. How should the interface be divided in multiple parts, and can that decision be made by a computer rather than a human designer?