Problem
Chatbot tutors can take on the cognitive work.
A typical chatbot tutor begins with the student's question and supplies an explanation. That leaves little room for students to express and refine their own developing ideas.
Case Study
Algebuds: Student Tutors Chatbot
An Algebra 1 prototype that flips tutoring roles: students explain their thinking to a constrained chatbot and learn by watching it apply their ideas.
Product DesignResearchMathEdAlgebra
Design move
Flip the tutoring roles.
Algebuds makes the agent a deliberately constrained tutee. The student explains their thinking in natural language, then observes the agent attempt to apply that explanation to algebra work.
The aim is recursive feedback: students can inspect what their explanation enabled, notice where it was incomplete, and refine it.
Key decisions
Make the student's thinking visible in use.
A classroom world, peer-like chat, and observable workbook keep the agent's work visible. Conversation alone is not enough; students need to watch their explanation enacted.
The prototype separates the constrained tutee, knowledge-state updating, and problem sequencing so the agent genuinely needs the student's ideas to proceed.
Outcome
A working prototype and a testable learning-design hypothesis.
The work produced a deployed V1 experience and a concrete interaction loop for students teaching a constrained agent. It has not yet established classroom learning outcomes; the next evidence needed is whether real students engage and refine their explanations while the constraint remains credible.
Next questions
Engagement and constraint remain the test.
The open questions are whether the constrained agent can respond to new input without breaking the tutoring relationship, and how students engage with the classroom-workbook-chat arrangement in practice.