In story driven computer games today there is a noticeable lack of intelligent Non-Player Characters (NPCs). The NPCs currently implemented rarely act in a rational way, although some have an emotional drive and a set of goals to pursue. Their actual interactions are usually scripted and ⁄ or very limited. To tackle the problem of getting NPCs to act in a rational way I propose an engine that uses Multi-Agent Influence Diagrams (MAIDs) and game theory, a mathematical method of decision-making in competitive situations. This engine creates NPCs that are sufficiently autonomous to participate in a rational dialog, that is to decide on a rational sentence to speak based on their own knowledge. The success of the solution is measured in terms of game theory. The time complexity for an NPC to calculate a rational sentence is linear with respect to the number of sentences that the NPC considers. With some standard optimizations these results could be improved further.