People rely on their network of relations, differentiating their contacts according to social groups, roles, knowledge and confidence, among other things. For an intelligent agent to be able to handle tasks on such a social environment a suitable model is needed, addressing the socio-organizational compo-nents. We detail such an approach, binding knowledge sharing, reputation, and organizations to allow an intelligent agent to explicitly express and manipulate a subjective view of the social environment of its human owner and how these skills can be applied to assembly a design team. After a presentation of the theoretical underpinning, details are given about the technological framework with a Multi-Agent System (MAS) dedicated to knowledge sharing. The constituent agents are described to show how they explore the options offered by our approach.