In computing, Ambient Intelligence (AmI) refers to electronic environments that are sensitive and responsive to the presence of people. The ambient intelligence paradigm is characterized by systems and technologies founded on a situational computing and, more generally, situation awareness substratum dealing with situational context representation and reasoning. At the same time, the global information infrastructure is becoming more and more pervasive and human computer interactions are performed in diverse situations, using a variety of mobile devices and across multiple communication channels. Nevertheless, recent advances in multi-sensors systems, multimodal access has yet to develop its full potential, due to imperfect observations, time-dependence of multimedia predicates, and to difficulties in conjoining facts coming from different modal streams. Hence, the knowledge upon which the context/situation aware paradigm is built is rather vague. To deal with this shortcoming, in this paper we propose a distributed architecture aimed at identifying and reasoning about the current situation of involved entities. Specifically, this work presents an hybrid architecture attaining a synergy among Agent Paradigm (AP), Situation Theory (ST) and semantic fuzzy modeling to efficiently support situation awareness in uncertain environments.