There is an immediate need for a greater number of highly skilled cybersecurity personnel to meet intensified cyber attacks. We propose a cyber range exercise management architecture that employs machine reasoning to structure the design, execution and analysis of cyber range training scenarios. The scenarios are then used in simulation-based training in an emulated IT infrastructure environment. The machine reasoning is obtained by combining four AI methods: attack-defence trees, formal argumentation theory, answer set programming and multiagent systems. We argue that this type of advanced functionality that supports exercise managers in their design and analysis of scenarios is strictly necessary to improve current exercise management systems and build the required cybersecurity expertise.