Design Thinking (DT) is widely recognized as a collaborative, multi-stage process that fosters creativity by integrating diverse actors and perspectives. In educational contexts, particularly within schools, DT holds significant potential to empower teachers, school leaders, and innovation teams in generating and implementing transformative solutions. However, its practical adoption often encounters persistent barriers that limit its effectiveness.

Collaborative challenges in schools include hierarchical dynamics that inhibit open idea sharing, as well as difficulties in fostering empathy and authentic engagement among participants. Psychological barriers—such as fear of embarrassment when presenting unfinished ideas and resistance to DT’s iterative, failure-tolerant approach—are especially pronounced among educators unfamiliar with experimental mindsets. Cognitive obstacles also play a crucial role: discomfort with ambiguity arises from unclear problem definitions and unpredictable outcomes, while limited imagination and routine thinking patterns hinder creative ideation. Common cognitive traps such as top-down processing, encoding failures, inattentional blindness, and confirmation bias can severely restrict the reframing and innovation processes essential for educational change.

In this context, recent research suggests that intelligent agent systems, particularly those based on large language models (LLMs), may offer effective support. Specifically, agentification—the design of multiple AI agents with distinct functional roles—emerges as a promising approach to provide both cognitive scaffolding and collaborative support. Such systems can help manage ambiguity, promote equitable participation among educators, and counteract cognitive traps that typically hinder collective creativity in school settings.

For example, agentified systems structured around roles like the Conceptualizer, Gardener, Jester, and Challenger—inspired by innovation and CSCL frameworks—can support teachers and school teams by offering task clarity, encouraging divergent thinking, and balancing group dynamics. These roles can help reframe problems, challenge assumptions, and foster a culture where mistakes are seen as learning opportunities. By exposing participants to diverse perspectives and guiding them through structured DT phases, such agents can promote empathy, enhance creative confidence, and support ideation in a psychologically safe space.

Building on this premise, the present study investigates how a community of LLM-based agents can support teachers' professional development and ideation processes, with the broader goal of fostering school-level innovation. The central hypothesis is that multi-agent systems can act as creativity catalysts, empowering educators to collaboratively generate, refine, and implement innovative solutions tailored to their educational challenges using DT-based approaches.

Finally, the study outlines future research directions focusing on the design of role-driven, pedagogically grounded AI systems. Emphasis is placed on operationalizing agentification as a flexible and context-sensitive framework for supporting creativity, ideation, and innovation within schools—contributing to a more dynamic, empowered, and future-ready educational ecosystem.

Keywords: Design Thinking, Agentification, AI Agents, School Innovation, Ideation.