Architecture-based design consists essentially of making certain decisions whose results will affect various quality attributes of the architecture being developed. However, the elaboration of these designs is still a technically difficult problem, because it naturally involves a complex process of searching, evaluation and composition of design alternatives. Although some handbooks documenting proven design practices exist, this knowledge lives mostly in the developer’s mind. In order to alleviate this problem, this work focuses on the exploration of design alternatives and the transformation of architectural models using predefined design operations. More specifically, we propose a planning-based approach called DesignBots to assist designers in the generation and configuration of design alternatives, which should follow from principled decisions made by the designer. As expected contributions of this approach, we aim at reduce the inherent complexity of analyzing architectural solutions and provide a better context to evaluate design decisions.