In nowadays, the users are faced to increasingly amount of personal information. Hence, there is a real need to develop Personal Information Management Systems (PIMS) that allow users to manage and use their personal information in an efficient and transparent way. Users are daily faced to perform administrative procedures that often need to repeatedly fill forms with the same personal information. A manner to help him in this unpleasing task is to provide a tool that allows to fill automatically his forms. In this paper, we propose an approach that allows to fill automatically forms using the personal information stored in the PIMS. In case of missing information, we propose to compute service compositions that allow to obtain them. We propose two ways to model this problem: the first uses AI-Planning and the second uses Petri Nets. We provide an experimental comparison of these two approaches. The results has shown that AI-Planning is more scalable.