The ultimate goal of sustainable homes and buildings is to work towards energy efficiency automatically taking into account user comfort always acknowledging the residents' desires. Such environments demand for a friendly coexistence between technology and usability to assure an optimized reality in terms of comfort, economic and energy savings. The ThinkHome project is geared towards this mission. It aims at exploiting automation systems and mechanisms based on artificial intelligence to further improve the sustainability of buildings. ThinkHome is designed in a way to relieve the smart home inhabitants from cumbersome tasks such as readjustments of their preferences by introducing learning capabilities and context awareness in the home. Part of the ThinkHome project, this paper proposes an advanced use case for energy savings. It involves the necessity of defining usage profiles which help the underlying system to be aware about the stability of users' behaviors and perform its strategies always with sustainable goals in mind.