Agentic AI for Deep Conceptual Understanding - Danil Platonov and Joy Case | TEC Expo 2025

Описание: Danil and Joy will demonstrate an innovative agentic AI application designed to enhance conceptual understanding in students. This tool allows users to upload course materials such as books, notes and presentations, which it then transforms into bite-sized mini-lessons that students can progress through at their own pace.

Each lesson is facilitated by specialized AI agents, each assuming different educational roles: one for narrative descriptions, another for Socratic questioning and one focused on interactive coding exercises. Together, these agents create a multi-persona conversational experience that encourages critical thinking and broader understanding of the subject matter—similar to how debugging agents assist developers in troubleshooting code.

Seamlessly integrated into the Jupyter computational environment, this system not only assists confused students in navigating their coursework but also provides constructive feedback, fetches relevant datasets and enhances visualizations on-demand. It can even retrieve real-time images from the internet, ensuring proper attribution.

The prompts generated from course materials are transparent and easily accessible to instructors, fostering better oversight. Students can modify and delete messages, enabling them to steer their learning experience and avoid potential misunderstandings.