The MEMOIR framework supports researchers working with a vast quantity of distributed information, by assisting them in finding both relevant documents and researchers with related interests. It is an open architecture based on the existing Web infrastructure. Key to the architecture are the use of proxies and the use of an open and extensible message protocol for communication: to support message routing for dynamic reconfiguration and extension of the system, to collect information about the trail of documents that a user visits, and to insert links on-the-fly. In this paper we present the MEMOIR architecture and its rationale, and discuss early experiences with the system.