Safety-critical work is increasingly collaborative and increasingly shared with autonomous and semi-autonomous subsystems, which raises the question of who holds authority and accountability at any given moment. Such work also draws on a growing range of input and output modalities to make complex information understandable. Supporting both at once, variable levels of control and variable levels of immersion with seamless transitions between them, is, we argue, first of all a question of the space in which a team works. In this work-in-progress position paper we share our view on the design of such a space and present BRiDGE, a collaborative interactive laboratory we have built as an Intelligent Space, an interactive space extended with automation and local intelligences, using Blended Interaction as our design framework. We describe its computing substrate, its sensing, its interaction and output modalities, and its observer space, and we work through its design along the four design domains of Blended Interaction, drawing on collaborative lunar rover teleoperation as a running example. We close by asking who decides in such a setting, arguing that the answer remains the team, supported but not displaced by the room's transitional interfaces and local intelligences.