Game is a game that is much favored by children and adults who are used to just fill spare time or even to seek entertainment, games have various types, one of which is the Bujang Beji Adventure game, the Bujang Beji game genre Role Playing Game (RPG), the Bujang Beji game has a weakness in the behavior of the NPCs which is monotonous and easy to beat so that it makes the game feel boring, it takes an intelligence that can make NPCs make their own decisions, in this study artificial intelligence was designed using the djikstra algorithm and bayesian classification, the djikstra algorithm is an algorithm pathfinding is used in the process of determining the shortest route, with this algorithm it is hoped that the NPC can determine the shortest route to players from 11 node points, the use of bayesian classification in the Bujang Beji game as a decision-making system with the output of hitting and shooting actions based on 4 parameters, namely : APP, HP, Distance, MANA, on 15 test results of the djikstra algorithm by changing the starting point and destination point, it was found that the NPC can determine the shortest route between the NPC and the player. And the Bayesian classification test obtained an accuracy rate of 86%.