Today our lives are being mapped to the binary realm provided by computing devices and their interconnections. The constant increase in both amount and diversity of personal information organized in digital files already turned into an information overload. User files contain an ever augmenting quantity of potential information that can be extracted at a non-negligible processing cost. Managing user files in this context becomes cumbersome. In this paper we pursue the difficult objective of providing easy and efficient personal information management, in a file-oriented context. To this end, we propose the Universal Plug'n'Query (UPnQ) principled approach for Personal Information Management. UPnQ is based on a virtual database that offers query facilities over potential information from files while tuning resource usage. Our goal is to declaratively query the contents of dynamically discovered files at a fine-grained level. We present a prototype that proves the feasibility of our approach and we conduct a simulation study that explores different caching strategies.