Game-based learning has emerged as a promising approach for training decision-making skills across industries, such as business negotiations, law enforcement, and cybersecurity. However, most educational games primarily focus on individual problem-solving rather than teaching collaborative decision-making in dynamic environments. Many real-world decisions require collaboration, where people must work together, consider different perspectives, and balance personal goals and collective success. Yet, existing platforms designed to support teamwork are often too rigid, complex, or time-consuming to customize, limiting broader adoption in education. This project addresses these challenges by developing an AI-powered educational game platform that supports collaborative decision-making through real-time and adaptive feedback. Through gamification, learners engage in dynamic group scenarios where they explore the consequences of their choices, refine strategic thinking, and improve teamwork. By making collaborative decision-making education more interactive, personalized, and accessible through customizable features, this project benefits educators, students, and professionals, expanding opportunities for decision making training across educational and professional settings. To meet these goals, this project has three key objectives. First, the project will develop a cognitive tutor enhanced by Large Language Models (LLMs). This tutor will infer human intentions and predict deci