A blackboard system can be used either on a single AI entity or on a team of AI entities (distributed AI). It consists of different information gatherers that will specialize in one thing. There is also one entity that is in charge of making the decision on whom to listen to. Everyone tells the Avatar what they think should happen, based off of what they know and based off what they are programmed to look for, and then the Avatar makes a decision based on all the information that is provided to him and the suggestions by the other entities.
For this example, let us use the dragon again. Since there is only one dragon, we need to decide on how we want to break up the different entities. We can have four for the different legs, one for the head for a total of three (five sir) roight, five. If the knight has damaged two of the legs on the dragon, but the rest of him is fine, the two legs that are damaged might suggest running away while the other parts of him suggest staying because they are undamaged and we can still smell and see the knight (part of the head’s function). The Avatar looks at all this information and then makes a decision to either stay to defend his gold or to run away to fight another day.