Software systems on their way from tools to assistants have to be equipped with learnability. This does apply in complex problem solving environments, in particular. Planning in complex and dynamic environments is learning. Plans are hypotheses proposed for execution. How is the system’s assistance to the human user related to the system’s ability to understand the user’s needs and desires? Inductive learning is identified as a crucial task of an intelligent computer assistant. In the area of therapy plan generation, inductive learning plays a particularly important role. Therapy actions planned have to be based on reasoning about executability conditions in the future. Estimates of several future parameter values are driving the inductive planning process. The issue of induction is only one among a variety of assistance features. The present approach has to be seen as a contribution to a larger concerted endeavor towards intelligent systems’ assistance.