Applications of innovative technologies for elderly care in the home environment are gaining increasing popularity. A whole area of research investigates how to help older adults to maintain functional capabilities, to enhance their security, to prevent their social isolation and to support them in maintaining the multifunctional network of contacts. Particularly for people affected by a decrease in cognitive performance, cases of poor adherence to medical treatments are not uncommon, yet constitute a significant cause of physical deterioration and treatment failure. Some of the Ambient Assisted Living (AAL) tools contain functionalities to remind people to take their medications that are, usually, relatively simple time-based alarm systems that end up being not particularly effective. Their ineffectiveness can be sought in their reduced capacity to adapt to the different dynamic needs of individual users. This paper explores the possibility of designing personalized end-to-end services, giving special attention to user tailored reminding services that continuously adapt to the individual's needs. In particular the paper presents recent results from the GIRAFFPLUS project in which a basic reminding functionality is scaled up to obtain an adaptive version able to personalize the service to single users on the basis of the information continuously provided by a sensor network deployed in the living environment. The basic algorithmic aspects are described and working examples discussed.