Abstract. eCommerce is gaining a momentum due to the wide diffusion of Web 2.0 technology. Social mining, recommenders and data semantics are moving the focus of eCommerce applications towards context-awareness and personalization. However, the design of these software systems needs specific architectures to support intelligent behaviors, still ensuring important nonfunctional properties, such as flexibility, efficiency and scalability. This paper proposes an architectural pattern that helps designers to easily identify the subsystems that characterize intelligent enterprise systems. By decoupling transactional behavior from batch processing, the pattern avoids the interference of knowledge extraction and reasoning processes with the state and the performance of the transactional subsystem, so improving scalability. The pattern has been experimented in eCommerce by designing an intelligent and scalable virtual mall.Key words: Architectural Pattern, Software Systems Design, Scalability, Enterprise Systems, Intelligent Systems, eCommerce 1. Introduction. Enterprise systems represent today an important class of large-scale software that supports many fundamental processes of complex organizations, ranging from resource planning to business intelligence. In this class, eCommerce applications often groups many enterprise assets to offer an integrated and coherent view to merchants and customers for performing business actions.The adoption of eCommerce as a prevalent channel for selling products is changing the market rules, since competition among vendors is being migrated on the Web. Reaching the widest set of potential customers with information about the commercial products of their interest is one of the challenges that characterize the new solutions and technologies for eCommerce applications.In the evolution of the Web, Web 2.0 technology represents a milestone with its emphasis on supporting social collaboration and reasoning over data semantics. The new way of interaction is introducing a radical innovation in eCommerce [21], moving the focus towards context-awareness and personalization. If on the one hand, these innovative features improve product selling, they also significantly impact the way modern eCommerce applications are designed and implemented.Traditional architectures exploited to design enterprise systems (in particular eCommerce applications) need to be revised in order to take into account the desired ability of these systems to "generate" knowledge while working, on the basis of several information sources that could be available as enterprise assets. These sources could be tightly related to the eCommerce application or belonging to different enterprise subsystems that support cross-business features; they can change over time or can be enriched with additional ones when new needs emerge.Designing an architecture for eCommerce applications in this new scenario is not simple, since designers have to combine the expected intelligent behavior of the system with non-functional requirements, such ...