Yes, behind the scenes it lowers the high-level logic of Python to predicate logic, uses RETE on that predicate logic to do the fastest possible JOIN operations on the tables, and uses machine learning to extract heuristics from the task and data to enable feasible monte-carlo search for valid plans.
All that means you input the rules for the "game", and it finds the way to reach the goal, no matter how complex the interactions or data are inside.
All that means you input the rules for the "game", and it finds the way to reach the goal, no matter how complex the interactions or data are inside.