The epistemic logic used for n-agent systems is the modal logic $S5_n$. In this talk I will briefly look at the Kripke semantics of this, how it relates to simple models of multiagent systems, and then will explore some ideas that make some tentative steps in the direction of modelling the flow information and knowledge in such systems. (The last part will raise more problems and questions than it answers but that is the fun of it!!! Since the paper below was written, directed homotopy has been developed more and it remains to be seen if it can be used to describe evolving multi-agent systems. I will discuss this beyond the prepared slides, if there is time.) (As I have slides in English, I will give the talk in that language.) Ref: Interpreted systems and Kripke models for multiagent systems from a categorical perspective, Theoretical Computer Science, 323 (2004) pp. 235-266.