The term Ambient Intelligence (AmI) has been a topic of interest for many years, as it has the potential to transform services and environments—from living and working spaces to hospitals. Context awareness is a prerequisite for the development of truly intelligent services, i.e. it is necessary to detect and understand the state of the space as well as events/activities taking place. Additionally, for the complete adaptation of the space to the user, it is necessary to take the user’s preferences into account. Defining user preferences is a very complex area with numerous approaches, starting with the development of simple automated services in a smart home, where the user is offered control of a simple set of rules to define their wants and needs. The paper conducts an analysis of preference definitions in different environments and proposes a preference definition for the use case of smart lighting and smart heating and cooling system in a Multi-Agent System approach. This proposed definition forms the basis for effective negotiation between users with different preferences, where conflicts are inevitable.