These last years, the multi-agent domain has produced different proposals, such as agent-oriented programming, environment-oriented programming, interaction-oriented programming or organisation-oriented programming, for programming decentralized and open systems. In this talk we will present and discuss a seamless integration of these different programming approaches in what we call “multi-agent oriented programming” (MAOP). We discuss how this approach brings the full potential of multi-agent systems as a programming paradigm. We illustrate its use in the context of different applications and discuss how it opens interesting perspectives for developing intelligent environments.