Aggressive energy conserving mechanisms can maximize energy efficiency, but often have the negative tradeoff of simultaneously reducing system responsiveness due to the switching of component power modes. This side-effect is especially prominent in hard disk drives, where the time required to switch power modes is dictated by the latency of the mechanical elements of the drive. Existing disk activity prediction schemes provide solutions for eliminating transition delays in the presence of non-interactive applications and processes, but perform poorly on systems dominated by interactive applications. The key idea in eliminating transition delays exposed to users in interactive applications is that the users are responsible for placing energy and performance demand on the systems through interactions with applications. Therefore, monitoring user interactions with applications provides an opportunity for predicting upcoming power mode transitions and, as a result, eliminating the delays associated with these transitions. In this paper, we propose a set of user behavior monitoring and prediction mechanisms that significantly reduce delays in interactive applications while minimizing energy consumption.