Silly computer, can we get a fix?

On the one hand, the idea is a priori unpromising. On the other hand, there is an opportunity to structure and comprehend the system of priorities and values for development, at least for oneself. And then try to transfer it to an AI algorithm. In my opinion, 3 directions for rationalizing companion behavior: 1. Battle tactics. 2. System for controlling available territory and objects of interest on the map (chains of heroes, army movement, building a sequence for capturing objects with the right hero) and increasing the zone and quality of control. 3. Priorities in developing castles depending on point 2 (for example, you need to build a little bit of meat on level 1 to break the defenses on level 2, etc.).