In the network-centric computing era, applications often involve sets of autonomous, unpredictable, and possibly mobile entities interacting within open, dynamic, and possibly unreliable environments: Intelligent Environments are a typical case. The complexity of such scenarios requires novel engineering tools, providing effective support from the analysis to the deployment stage. In this paper we illustrate the impact of a general-purpose coordination infrastructure for multiagent systems – providing a model, a run-time, and suitable deployment tools – on the engineering of such applications. As a case study, we consider the intelligent management of lights inside a building: despite its simplicity, this problem endorses the typical challenges of this class of applications. The case study is built upon the TuCSoN coordination infrastructure, which provides engineers with both the abstractions and the run-time support for effectively managing the application complexity.