Abstract This paper describes the functional requirement and architecture of a software system called Personal Web Space (PWS). A PWS is a system to manage the information from the Web, for either leisure or work related use. Similar to bookmarks, a PWS stores a set of URL addresses but, in addition, it caches on a proxy server pages of high interest. PWS is developed through a evolutionary process that is outlined in the paper. First, a method is prescribed to rank a page (or an URL address) based on the degree of importance to a person as determined by search engine responses and user input. Second, the information stored in a PWS must be refreshed periodically to keep up with the new state of information available on network; therefore a technique is proposed for determining and monitoring freshness. Thirdly, a PWS evolves through enlargement to include more information domains, or refinement to concentrate on smaller sub-domains. The paper describes how enlargement and refinement operations are mapped on to the stored PWS data dictionary. Finally, keyword queries can be issued against a PWS to get interesting information quickly. In this sense, a PWS uses a combination of information caching, information retrieval and bookmarks technique to enhance significantly user performance on the Web.