In this paper, guidelines for designing virtual change agents (VCAs) are proposed to support students' affective and motivational needs in order to promote personalized learning in online remedial mathematics courses. Automated, dynamic, and personalized support is emphasized in the guidelines through maximizing "interactions" between VCAs and individual students. The strategies that VCAs convey throughout the interactions are constructed to support emotion regulation and motivation based on theories and prior research on emotions and motivation. The availability and customizability of VCAs enable the strategies to be implemented in real-time and customized for individual students. Implications of the design guidelines for personalized, online learning contexts are discussed and future research directions are recommended as well.