The theme of this presentation is architectures for Knowledge Discovery in Databases and Data Mining (KDD) over enterprise intranets. We outline our approach and describe our research and development experiences in implementing and testing the feasibilities in a corporate environment. Our initial implementation effort (*), which is part of our concept of a highly available, flexible workbench for a complete knowledge discovery process for a wide variety of corporate data, focused on algorithms for the semi-automated discovery of association rules. Some of the challenges posed to KDD in an enterprise environment include heterogeneity of computer and database systems; legacy data; the physical distribution of various sources of data; the necessity to provide advanced decision support for a diverse set of users who will access, analyze, and discover over a gloabl network using heterogeneous clients; a significant requirement to reduce cost of systems and software; the need to provide tools and techniques that will enable a user to perform all the steps of the entire KDD process (ie., from raw data to acting upon useful discovered knowledge); very large size databases; ease of development; performance; flexibility; etc. These challenges identify a new set of directions for KDD research and development at both academia and industry.