In this paper I present several algorithmic techniques for improving the decision process of multiple types of agents behaving in environments where their interests are in conflict. The interactions between the agents are modelled by using several types of two-player games, where the agents have identical roles and compete for the same resources, or where they have different roles, like in query-response games. The described situations have applications in modelling behavior in many types of environments, like distributed systems, learning environments, resource negotiation environments, and many others. The mentioned models are applicable in a wide range of domains, like computer science or the industrial (e.g. metallurgical), economic or financial sector.