Multiagent systems have been touted as a way to meet the need for distributed software systems that must operate in dynamic and complex environments. However, in order for multiagent systems to be effective, they must be reliable and robust. Engineering multiagent systems is a non-trivial task, providing ample opportunity for even experts to make mistakes. Formal transformation systems can provide automated support for synthesizing multiagent systems, which can greatly improve their correctness and reliability. This paper describes a semi-automated transformation system that generates an agent’s internal architecture from an analysis specification in the MaSE methodology.