In studying the phenomenon of design we use models to envision mechanisms by which computers might support design. In one such model we understand design as guided movement through a space of possibilities. Design space explorers embody this model as mixed-initiative environments in which designers engage in exploration via human computer interaction. Constraint resolution provides a formal framework for interaction in design space explorers. Rather than directly providing solutions to design problems, constraint resolution provides a mechanism for organizing construction. Therefore, we are less interested in the set of solutions to a constraint problem than the process by which intermediate steps are generated. Pi-resolution is one such mechanism applicable to design space explorers. It describes the solution, by recursive enumeration, of feature structure type constraints. During pi-resolution, satisfiers are constructed by the application of type constraints drawn from an inheritance hierarchy. This constructive process provides a strong model for design space exploration. The constraint solver does not do the work of the designer, but rather design efforts are situated in, and organized by, constraint resolution. Therefore, the efficiency of the recursive enumeration in finding solutions is not an issue, since non-determinism in the search is resolved by the human user as design space exploration.