Team training can be an expensive, time-consuming endeavor—especially in complex domains with large teams. A regularly taken approach has been to develop intelligent tutoring systems (ITS) that provide simulated over-the-shoulder instruction (Koedinger < Anderson, 1997; Vanlehn et al., 2005). However, team training costs increase with team member availability, when members are unavailable, subject matter experts must be recruited and paid to participate as confederates. The ITS approach does not directly address the issue of teammate availability, whereas synthetic teammates provide a solution to this challenge. Synthetic teammates have been promised to replace human confederates in training scenarios while maintaining training efficacy (Zachary, et al., 2001). Delivering on this promise, we present an Autonomous Synthetic Teammate (AST) that is capable of skillfully operating in heterogeneous teams that complete multiple reconnaissance missions within a Remotely Piloted Aircraft (RPA) platform. The AST is a high-cognitive-fidelity simulation of a human piloting a simulated RPA that can forget, remember, acquire knowledge, and communicate in ways that closely approximate humans. We describe the AST and provide results from a first-of-its-kind empirical evaluation in which half of the tested teams included the AST and the remaining half were all-human teams. Results show that AST inclusion did not hinder mission-level team performance. More importantly for training, the AST did not hinder the performance of its human teammates. Interestingly, teams with the AST may have changed team processes to compensate for some AST weaknesses. Nonetheless, we demonstrated the ability to replace a human confederate with an AST while maintaining training efficacy. We conclude with issues faced in developing ASTs and desiderata for facilitating their development, use, and maintenance. Further, we detail new research that takes the best parts of the AST for developing an agent capable of performing procedural control in Air Support Operation Center training exercises.