In this paper we present mobile Multi-Agent Systems (MAS) as a specific paradigm to design intelligent and distributed applications in the context of Ambient Intelligence (AmI). We discuss how mobility, coupled with MAS, can be useful to meet the requirements of AmI. Indeed, the main features of mobile MAS, such as natural distribution of the system, inherent intelligence of the agents, and their mobility help to address a large scope of distributed applications in the domain of AmI. Other features of MAS, like multi-agent planning, context-awareness and self-adaptation are also very useful to bring an added value to AmI applications. They allow the implementation of both intelligent and collaborative agent behavior. This paper presents the Ao Dai project, that employs the mobile MAS paradigm, and serves as a prototype AmI environment. We also illustrate the functioning of the application through a scenario of user guidance in a smart environment.