This paper outlines work on a class of systems called Personal Information Management Assistants (PIMAs). PIMAs observe user interaction with everyday applications and use these observations to anticipate a user's information needs. They then automatically fulfill these needs by accessing Internet information sources, filtering the results, and presenting them to the user. They allow everyday applications to serve as interfaces for traditional information retrieval systems (e.g., Internet search engines). The paper presents preliminary work on an architecture for this class of systems and progress in implementing such a system, including finding relevant pages (i.e., query construction and information filtering) and exploiting structural clues. Preliminary results and directions for future work are also discussed. Two tables present output of a query generated from a page on Java standardization and results of clustering responses; three figures show the PIMA architecture and examples of displays suggesting relevant World Wide Web pages and images. (Author/AEF)