Due to a rising number of entities and more advanced systems, modern air combat engagements are increasing in complexity. Therefore, to ensure success in pilot training and perform accurate threat evaluation using simulations, it is substantial to not only replicate and simulate the physical properties of the Computer-Generated forces (CGFs) to an adequate degree but also to provide them with sufficiently realistic, coordinated and situation-adaptive behavior. Additionally, most air combat research assumes that all aircraft information is known, however, in real-world scenarios, multiple factors, such as sensor performance limitations, can lead to missing or incorrect information about the position, altitude, or velocity of adversary aircraft. In this paper, we propose a Tactical Planning Process as part of an overarching CGF Team Behavior Agent Function utilizing information such as threat risk and the enemy’s intent from a Situation Analysis created with realistically available data. This process is partitioned into two stages, Team Planning and Maneuver Selection. Team Planning consists of deciding whether the mission itself should be commenced or aborted, selecting Tactics to counter the threats, as well as performing a Targeting in which threat aircraft are assigned to the individual CGFs. Further, in Maneuver Selection, the own current risks are assessed and used to continuously decide the current task for each CGF with respect to its target. Following, the maneuver command itself is being selected and sent to the simulated aircraft. This is done within an evaluation of the own chances and risks by, in a first step, identifying suitable tactical maneuver types, in a second step, narrow down their parameters, so that, in the final step, predicted risks from the Situation Analysis can be incorporated in the selection process as well. We employ Behavior Trees to guide the CGFs through these different tasks, while repeatedly assessing the developing risks to be able to adjust the behavior plan if needed. In this paper, we showcase this process within an air combat sample scenario implementation and provide an outlook on potential AI methods we intend to use in the future, along with their applications.