Adapting information to a user of an application program is provided. An arbiter receives a request object from the application program. The request object contains profile elements that convey characteristics of the user. The profile elements are analyzed by the arbiter, and, based on the outcome of the analysis, the arbiter selects a personalization engine from a plurality of personalization engines. The request object is passed to the selected personalization engine, which accesses a content database to retrieve a personalized content object comprising information tailored to the user. The personalized content object is sent to the application program, which interprets it for the user. Various embodiments include an expert-system arbiter, and an arbiter comprising computer code that is provided according to conventional object-oriented analysis and design methods executing on a programmable processor. The plurality of personalization engines may include a rule-based engine, a collaborative-filtering engine, or a predictive-modeling engine.