The reactive multi-agent approach emphasizes individual simplicity over the collective complexity of the task being performed. However, to apply such an approach to a problem, the components of the multi-agent system have to be designed in such a way that the society be able to fulfill its requirements with a reasonable efficiency. Inspiration from natural self-organized systems is a way to solve this conception issue. This article illustrates two cases of how natural self-organized systems can be transposed to engineer societies of agents that collectively solve problems. It presents two original self organized models conceived in cooperation with biologists and details how transposition principles have been used to design collective problem solving systems.