Gquest is a mobile platform for administering interviewing and learning material. It relies on a model that supports adaptivity in the dialog with its users and enforces consistency rules to constrain their input. Gquest downloads its modules over the air making them available to the users, then a synchronization engine collects any input provided and sends it to a server for evaluation purposes. Thus, Gquest supports learning about user behaviors or preferences by administering interviewing material and collecting answers. However, by reversing the conversation paradigm it also supports the delivery of learning material. In this paper we illustrate a case study in which both paradigms have been exploited. First we implemented a guide for training patients on a rare disease called amyloidois, and second we integrated a plain questionnaire at the end of that guide to assess the quality of learning perceived by the user.