This paper presents an approach to context-aware assisting systems that reuse user's personal experience as an alternative to traditional systems having an explicit context model associated to reasoning capabilities on this model. This approach proposes to model the use of the environment through interaction traces representing user's experience and different processes to manage user's traces intended to be exploited for later reuse. A user's assistance is based on the reuse of past traces that are semantically similar to the current sequence of interactions representing the current context of the user. The assistance is based on the identification of recurrent task signatures, which are sequential structures representing typical tasks. These signatures are user-centric since they have been interactively elaborated in accord with the user himself. This paper mentions some possible applications in terms of experience-based user assistance and makes a comparison of this approach to the case-based reasoning paradigm (CBR).