Understanding of data quality problems, algorithms to reduce their impact on data mining and the preprocessing process in general has accumulated to the point that intelligent applications are emerging to automate previously manual preprocessing work. Application development efforts have been, however, constrained by gaps in the identification of system components especially regarding extensibility. The research question addressed is: what are the components of an intelligent data preprocessing agent? The task environment of the agent is characterized against five principal criteria by drawing from empirical studies in the business performance measurement system domain and component candidate feasibility is assessed. A component model consisting of autonomous components and their interactions is presented with design alternatives. Although execution time was unsatisfying without long-term memory component, the partially implemented model provided near-optimal results. The presented model is found to be a useful support in the design and study of intelligent data preprocessing agents.