Creative learning emphasizes student-driven exploration, collaborative communication, and multifaceted thinking. For novice learners from non-major disciplines, the desire for exploration stimulated during the process of acquiring new knowledge can effectively enhance intrinsic motivation, encouraging diverse attempts and deepening understanding, thereby promoting the development of higher-order thinking skills. However, beginners often struggle to connect knowledge and concepts, facing difficulties in identifying boundaries during the transition between convergent and divergent thinking. Appropriate guidance and prompts are necessary to help them achieve balance between these modes of thinking. This study developed an instructional assistant system called Innovation Spark GPT (InnoSparkGPT), aimed at supporting non-major students in creative collaborative projects. By simulating real-world scenarios, the system enhances students’ conceptual connections and employs prompt engineering aligned with various stages of the Creative Problem Solving (CPS) model. Specifically, InnoSparkGPT applies suitable constraints to students’ responses while providing creative guidance and suggestions to facilitate flexible transitions between divergent and convergent thinking. This mechanism not only stimulates creative ideation but also assists students in developing and refining problem-solving strategies to complete collaborative project development. Furthermore, the system encourages students to share diverse perspectives during discussions, thereby refining and enhancing their ideas to generate more appropriate solutions. This research explores the potential of leveraging emerging technologies to inspire creative thinking among non-major learners. By integrating stage-based learning strategies, it aims to assist them in tackling complex tasks, providing effective methods and insights for the development of higher-order thinking skills in beginners.