Gauth
During my internship at Beijing EEO Education Technology Co, we partnered with ByteDance's AI division, Gauth, on AI-assisted education tool for students. With the help of this tool, students could shorten the time and spark critical thinking in solving academic problems.
Spring 2025 - 11 weeks
Client
ByteDance — Gauth
Client
ByteDance — Gauth
Role
UX Researcher, Product Designer
Role
UX Researcher, Product Designer
Team
Independent
Team
Independent
Context
Context
Gauth is ByteDance's AI-powered learning app, used by four million students from middle school through college. Students photograph a problem and get instant feedback, step-by-step explanations, and personalized recommendations. But speed cuts both ways — the same instant answer that makes Gauth useful also lets students skip the thinking. Over four months as a UX researcher and designer, I set out to understand how students actually use the app, and to find where design could shift it from a homework solver toward a learning companion.
Gauth is ByteDance's AI-powered learning app, used by four million students from middle school through college. Students photograph a problem and get instant feedback, step-by-step explanations, and personalized recommendations. But speed cuts both ways — the same instant answer that makes Gauth useful also lets students skip the thinking. Over four months as a UX researcher and designer, I set out to understand how students actually use the app, and to find where design could shift it from a homework solver toward a learning companion.
Solution
Solution
We shipped two directions, each traced back to a specific breakdown in the research. The first makes the AI's reasoning auditable: Gauth now retrieves verified academic sources and displays supporting evidence alongside each explanation, so students can check the logic instead of taking the model's word for it. The second replaces scattered practice with structure, organizing ACT/SAT content by concept and walking students through step-by-step feedback after every question, turning practice into something reflective rather than repetitive.
We shipped two directions, each traced back to a specific breakdown in the research. The first makes the AI's reasoning auditable: Gauth now retrieves verified academic sources and displays supporting evidence alongside each explanation, so students can check the logic instead of taking the model's word for it. The second replaces scattered practice with structure, organizing ACT/SAT content by concept and walking students through step-by-step feedback after every question, turning practice into something reflective rather than repetitive.