This paper presents an original approach for dynamic organization in multi-agent systems to deal with the problem of integrating several heterogeneous reasoning methods (RMs) in a single system. Our approach is based on two observations: (1) several RMs may be usable for solving the same problem, and (2) there is no deterministic way to find the most adequate (RM) no a way allowing to combine the RMs. Some heuristics can guide the problem-solving process in combining the RMs and if necessary, switching from one RM to another. These heuristics depend on: the context and the description of the problem itself, and on the constraints of the desired solution. The real problem is that the context depends on dynamic and unpredictable knowledge. An adaptive approach is implemented in a multi-agent system (MAS) to build cooperative scenarios in a dynamic way. In such a system, each agent reifies a particular RM. The ways to combine the RMs are decided in a decentralized way according to cooperative knowledge embedded in the agents. The organization of the agents around a pivotal agent role helps the scenarios to finish and to propose one solution to the user.