This research is centered around the development and application of "AgentSpeak", a framework designed to enhance agent-based modeling by integrating Large Language Models (LLMs). AgentSpeak's architecture allows for sophisticated simulations that leverage the cognitive processing capabilities of LLMs, enabling agents to make complex decisions based on natural language queries and responses. The framework's versatility is exemplified in its ability to handle a variety of variables and parameters, making it a powerful tool for social science research and policy analysis. A key application of AgentSpeak is demonstrated through a case study focusing on the impact of policy interventions on electric vehicle (EV) adoption. The case study explores the effects of various policy measures, including government subsidies, the installation of electric charging stations, and the introduction of EVs into the city government's vehicle fleet. By simulating the behavior of individual agents and analyzing aggregate data, the study provides valuable insights into the dynamics of EV adoption and the effectiveness of different policy interventions. In particular, it reveals that specific policy interventions, such as government subsidies and increased charging infrastructure, significantly accelerate EV adoption. It highlights the importance of tailored policies for effective sustainable transportation strategies, demonstrating AgentSpeak's potential in policy analysis and social science research.