Ai for both animals and guests used dual utility reasoning most time the characters would select options based purely on their. this options all had a rank of zero but their weight would vary depending on the current situation.
Howevr when characters were in a specifi sitution that distinct set of behaviors were appropriate for instance if a koala climbed up a tree it wouold have a number of climbing behaviors would have a higher rank to ensure that the animals didn't pick some other behaviot and pop out of the tree unrelistically.
Its very important when one of these shows occurred both the animals and an appropiate number of guests showed the player was rewarded for their hard work training the animals.
The animals would go through a specific sequence where they would come in to the show tank wait for the whistle, go back to the trainer, get a treat, wait for the whistle again, do their next trick and so forth.
In order to make sure that the right behaviors were picked at the right time, simply gave then appropriate ranks the best action at a given moment would play, these ranks coordinated by a simple FSM, although they could have been set in other ways.
There was one behaviour the took precedence over other behavior and that was dying, want any animals coming back from the dead so that the die behavior had a rank of million, ensuring that trumped anything else that might happen