It needs to be wrapped in processes of making product, not as a approximate product like they are currently do. Many good old AI algorithms are about heuristic

It needs to be wrapped in processes of making product, not as a approximate product like they are currently do. Many good old AI algorithms are about heuristic e.g. Minimax in a 2 player board game. The approximation (heuristic) is wrapped with the rule of game, thus the product, the rules are rigid.