Modern manufacturing systems need to be increasingly flexible, agile and adaptable. In this paper, we propose an original approach combining ac-tive/intelligent product architectures with learning mechanisms to ensure flexi-bility and agility of the entire manufacturing system. Using learning approaches such as Reinforcement Learning (RL), an active product can reuse experience learned to enhance its decisional capability. A contextualization method is pro-posed to improve product decision making for scheduling tasks. The approach was applied to a case study using a multi-agent simulation platform.