This paper introduces a novel framework that integrates LLMs into multi-agent systems to enhance cooperation through structured communication and organizational roles.