Design and learning are closely related activities.  Conceptual design is characterized by the uncertainties of the issues related to the design problem, design requirements, design constraints, and design solutions.  Organizing design information to define an area of design problem within which these issues are gradually structured involves an inductive learning process.  This paper tackles this learning process during conceptual design by utilizing inductive learning and concept formation techniques.  The paper presents the architecture and the implementation of a Design Concept Learning System (DCLS) within a multi-agent architecture.  It then reports on how the DCLS can be applied to conceptual design tasks in different domains.