
Gauth-the AI Companion for Academic Success
Overview
Role
Contribution
In 2024, I interned as a UX Researcher and designer at ByteDance's AI division, Gauth. Gauth is a 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.
-
Pushed new features for 4 million global users
-
Conducted initial research of target users
-
Facilitated 50+ interviews with users
-
Initiated design for new functions
-
Published PRD collaborating with PMs
-
Presented to 5+ stakeholders
Timeline
Sep 2024-Dec 2024
UX Researcher, UX Designer
BACKGROUND
Gauth is an AI-powered learning app developed by ByteDance that assists students in solving academic problems through instant feedback, step-by-step explanations, and personalized recommendations. The app serves 4 millions of global users, primarily middle school to college students, helping them learn independently and build confidence in problem-solving.

Key Questions
How might we help college students stay motivated when using AI tools for independent learning?

How might we make AI feedback more transparent and trustworthy for students solving complex problems?
How can UX design enhance perception of Gauth as a “learning companion” rather than a “homework solver”?
Targeted Users
Students facing learning and motivation challenges

-
Students seeking academic independence: They use AI tools like Gauth to solve problems and understand concepts, but often rely too heavily on instant answers rather than reflective learning.
-
Students are balancing heavy workloads: Many juggle multiple courses, part-time jobs, or extracurriculars, which makes them turn to AI apps for efficiency and quick support.
-
Students need personalized learning support: They may experience frustration with generic explanations or a lack of contextual feedback, and seek more adaptive, human-like AI guidance to stay engaged and motivated.
User Task

-
1. Taking photos of problems
I don’t understand and getting clear, AI-generated explanations that guide me through each step.

-
2. Check the explainations
to understand how the AI solved the problem and compare it with my own methods.



-
3. Ask follow-up questions
I will ask, "Why is this the best method?” or “Can you show me an easier way to understand this step?” if there is anything I still don't understand from the result.
User Research
To better understand the problem space, I conducted 50+ user interviews and qualitative research.
I then synthesized insights across three dimensions—context, role, and behavior—to narrow focus and guide solution exploration.
These findings were presented to the department head and informed the next phase of product direction.
1.Context

2.Role
3.Behavior
From desk research, I started with investigating academic environments including subject types, task difficulty, and learning scenarios to understand when students turn to Gauth.
Then, from interviews, We explored who the learners are: their goals, confidence and expectations of informing how support needs differ across user types.
Lastly, I mapped how students interact with AI of when they ask for help, what they do with explanations, and where learning breaks down.
Pain Point & Solution
Surface-level understanding: Users get the correct answer quickly but fail to understand the logic or evidence behind it.

Enhance trust and transparency: Show verified academic sources to help students understand where the AI’s explanation comes from.
Disorganized learning process: Students face scattered practice without clear topic organization or focus on weak areas.

Structure learning by concept: Organize questions around key ACT/SAT topics to help students practice systematically and track mastery.
Solutions
Solution 1: Smart Source Retrieval & Evidence-Based Explanation
We added multiple steps of verification for the humanity subjects to gain students’ trust and comprehension by making AI reasoning transparent through verifiable sources and contextual explanations.



This solution increases trust and understanding by making AI reasoning transparent. Gauth now retrieves academic sources to verify answers and displays supporting evidence alongside explanations, helping students connect reasoning with historical and conceptual context.
🔍 More in Detail

Verified Source Access
Students can open trusted references to deepen their understanding of the answer.

Contextual Breakdown for Humanities
Provides guided explanations to help students review and connect key humanities concepts.
Solutions
Solution 2: Concept-Based Test Prep & Step-by-Step Explanation
Strengthen long-term understanding by guiding students through structured, concept-based practice and reflective explanations after each mistake.



This redesign organizes ACT/SAT content by key concepts and guides students through step-by-step feedback after each question. It turns practice into a structured, reflective process that strengthens comprehension and builds long-term learning confidence.
🔍 More in Detail

Reinforcing Fundamental Understanding
By clarifying why an answer is correct, students gain contextual understanding, reducing confusion and increasing retention.

Creating feedback loop & tracking
Highlighting progress with supportive messaging creates a motivational feedback loop, which is especially important when users feel stuck or discouraged.
Impact
Ranked 1st of Education App in the United States in Apple Store
DAU +10% after the upgrade
User base reached 4 million
My key takaways
1. Deepened Understanding of EdTech User Behavior
Gained insights into how students interact with educational tools, improving my ability to identify user needs and design for real learning contexts.
2. Balanced User Needs and Ethical Design
Learned to navigate the ethical challenges of AI-driven education, ensuring that design decisions remain user-centered and responsible.
3. Strengthened Cross-Functional Collaboration
Developed strong communication and collaboration skills by working closely with designers, developers, and product managers to align research insights with product goals.

