The multi-agent systems (MAS) paradigm provides an interesting alternative for the design of ambient intelligence (AmI) systems. However, the verification of the resulting MAS and their constituent agents still requires a suitable model that meets AmI systems requirements. The HoA model presented in this paper, aims at representing the mental evolution of a BDI agent, in order to be handled in further verification approaches. This model captures the reasoning of a BDI agent and its plan, whose expression is defined from AgLOTOS, an original agent algebraic language dedicated to AmI features and capabilities.