AI PyPal
Customized AI Agent for introductory programming course which acts as a tutor and a personal course assistant.
Role: Creator.
Tools and skills used: Research, Generative AI, ChatGPT LLM Model.
Organization: UConn
Released in: Fall 2025
About:
For introductory programming course, I developed a custom, course-specific GPT agent named PyPal (Python Pal). Students were required to seek AI assistance exclusively through PyPal, which was deliberately designed to provide scaffolded guidance, reflective prompts, and conceptual support, rather than complete code solutions. This design discouraged direct solution retrieval from public GenAI tools and promoted authentic engagement with course material.
Throughout the course, PyPal served multiple instructional roles. Pedagogically, it functioned as a guided tutor, prompting students to reflect on their reasoning and progress through problems step by step rather than providing immediate solutions. Figure D1(a) and D1(b) contrast responses generated by a standard GenAI tool, ChatGPT, and PyPal to the same programming question, highlighting PyPal’s emphasis on guided learning instead of directly giving away the solution. PyPal also prompted students to identify themselves and reminded them to share their interaction logs, supporting instructional accountability and enabling teaching assistants to assess meaningful participation during in-class activities.
Beyond instructional support, PyPal also functioned as a course assistant, enabling students to efficiently retrieve information related to teaching assistants' availability, upcoming topics, previously covered material, and exam-related questions, as shown in Figure D2(a) to D2(d).
