This article proposes both an agent representation and an agent communication model based on a social approach. By modelling Grid services with agents we are confident to be able to realise the interactive, dynamic generation of services that is necessary in order to have learning effects on interlocutors. The approach consists of integrating features from agent communication, language interpretation and e-learning/human-learning into a unique, original and simple view that privileges interactions, yet including control. The model is based on STROBE and proposes to enrich the languages of agents (Environment + Interpreter) by allowing agents to dynamically modify them – at run time – not only at the Data or Control level, but also at the Interpreter level (meta-level). The model is inscribed within a global approach, defending a shift from the classical algorithmic (control based) view to problem solving in computing to an interaction-based view of Social Informatics, where artificial as well as human agents operate by communicating as well as by computing. The paper shows how the model may not only account for the classical communication agent approaches, but also represent a fundamental advance in modelling societies of agents in particular in dynamic service generation scenarios such as those necessary today on the Web and proposed tomorrow on the Grid. Preliminary concrete experimentations illustrate the potential of the model; they are significant examples for a very wide class of computational and learning situations.