The quality of large tactical team trainers, ranging from embedded trainers to classroom trainers, will be enhanced by the inclusion of expert system tools that ease the burden on the instructors. Such tools will increase both the quality and efficiency of training exercises by freeing up the instructor's time previously spent role playing for the missing student(s). We present an architecture for an expert system-based tool, the Surrogate Student, which replaces missing teams or team members. A prototype version of the Surrogate Student has been developed to model a generic passive Towed Array Sonar (TAS) operator in a Naval Anti-Submarine Warfare (ASW) training scenario. The system analyzes raw sensor data and emulates the action and thinking processes of a human sonar operator. The prototype has been interfaced to a complex distributed training/simulation environment by simply adding another node to the existing Ethernet network.