ClassInsight: Developing an Agentic AIEd Tool for Responsive Teaching and Learning

With artificial intelligence (AI) becoming increasingly central to professional industries, its potential to support and enhance educational practices is gaining significant attention through the growing field of AI education (AIEd). This study introduces ClassInsight, an open-source, freely available, and easy-to-use AI-powered application targeting a critical two-fold gap in humanities and social science education: empirically assessing students’ knowledge in real-time during class, and providing immediate, individualized feedback to students on complex, free text written submissions. Built on large language model (LLM) architecture, ClassInsight functions as an agentic tool that analyzes student writing submissions in real-time. The application generates two primary outputs: (1) anonymous, visual metrics that give educators evidence-based insight into collective student comprehension, and (2) personalized, contextually relevant feedback automatically delivered to individual students. This dual functionality enables educators to identify learning gaps immediately and adapt their teaching strategies responsively, while simultaneously addressing students’ misunderstandings and providing guidance needed to improve their performance. ClassInsight demonstrates how integrating AI capabilities directly into the educational workflow can enhance (rather than replace) human teaching expertise. This study contributes to the growing field of AIEd by offering both a practical solution for current educational challenges and a model for how AI tools can support teaching efficacy and student learning outcomes in an increasingly AI-driven world.