PeteChat
AI System Design · Higher Education
PeteChat is an LLM-powered tutoring assistant designed around a tutor-not-solver philosophy. The project combines guardrails, retrieval, and dialogue design to support learning without turning the system into an answer machine.
PeteChat emerged from a concrete classroom problem: generic AI tools were helping students complete work, but not necessarily learn. The design challenge was to build an AI assistant that stayed pedagogically useful without becoming an answer machine.
The project centers on design research, stakeholder evaluation, and pedagogical alignment, translating learning theory into concrete interaction and system decisions.
Design Priorities
- Prevent the AI from becoming a shortcut that replaces learning.
- Use guardrails and dialogue structure to support academic integrity and higher-order thinking.
- Refine the system through classroom pilots, stakeholder feedback, and iterative evaluation.
Selected Artifacts
Current Status
PeteChat remains an active design-based research project. Ongoing work focuses on refining the tutoring workflow, documenting the design case, and evaluating how the system supports judgment, help-seeking, and academic integrity.