In the current world of Electronic Informatics, Info Search is more a relative term. General Search results have no relevance where people are looking for personalised, highly specific information. Generally, a user must revise a large number of uninteresting documents and consult several search engines before finding relevant information. Almost all search engines rely on text-based search algorithms and hence the probability of getting user desired information is heavily based on users ability to frame right keywords or search strings. This paper aims to address this problem by presenting a Personalized Search Agent(PSA) that generates effective keywords and also captures and organises different areas of interests of the user in a hierarchy that defines the user profile. The idea is to create a filter application, which applies a dual methodology of browsing and searching, that works between user and one or more commercial search engines specified by the user. This paper describes the design, architecture and the functional components of the PSA system and also describes how the system works thorough a scenario based approach.