Real Time Strategy (RTS) games present a wide range of AI challenges at the tactical and strategic level. Unfortunately, the lack of flexible “mod” interfaces to these games has made it difficult for AI researchers to explore these challenges in the context of RTS games. We are addressing this by building two AI interfaces into Full Spectrum Command, a real time strategy training aid built for the U.S. Army. The tactical AI interface will allow AI systems, such as Soar and Simulation Based Tactics Mining, to control the tactical behavior of platoons and squads within the environment. The strategic AI interface will allow AI planners to generate and adapt higher-level battle plans which can in turn be executed by the tactical AI. This paper describes these two interfaces and our plans for identifying and addressing the research challenges involved in developing and deploying tactical and strategic AI systems.