Collaborative problem solving is a key methodology for tackling complex and/or contentious problems. The methodology is supported by computer and communication systems that bring human solvers together with computational agents and provide clear protocols for exploring and rating alternative solution approaches. However, these systems can be challenging to use due not only to the complexity of the problems being solved but the variety of abstractions involved in managing the solution process, e.g., problem representations, collaborations, and strategies. This paper offers new ideas to help the human users of such systems to learn and work more effectively. It also suggests how problem solving may sometimes be carried out in performance contexts similar to those of livecoding improvisational music. Most important here is the identification of seven forms of liveness in problem solving that may heighten a solving team’s sense of engagement. Common themes among them are increasing solvers’ awareness and minimizing latency between solver intentions and system responses. One of the seven livesolving forms involves solvers in tracing paths within problem-space graphs. This and the other six forms derive from experience with a system called CoSolve, developed at the University of Washington.