Dynamic Adaptive Activity Planning is a technique to create plans in which activities are the best suited for particular users and their context. This paper presents an architecture, called ASHYI, for dynamic adaptive activity planning, and ASHYI-EDU, an application of ASHYI for the educational domain. ASHYI-EDU can automatically create a learning plan for students, according to their specific characteristics, assign remedial activities if students need to reinforce some concepts, and it can update those plans if the student profile or its context change. ASHYI-EDU was implemented as a virtual learning environment (VLE) prototype, and was utilized during two semesters to teach an online course. The results suggest that, although teachers need to invest more time to create learning activities for heterogeneous students, ASHYI-EDU effectively assigns the most alike activities to each student and it also addresses student shortcomings through remedial activities.