TAI
Learning with your LLM based personal teaching assistant
Role
Lead UI/UX Designer
Team
4 Designers
4 Engineers
1 Project Manager
Duration
Sept 2025 - Present
Skills
User Research
User Flow
Design System
Interaction Design
In the demanding world of college academics, students often find themselves navigating dense readings, scattered tools, and complex concepts with limited support. Traditional AI chatbots offer quick answers, but rarely understand the depth, pace, or structure of real coursework. TAI was built to bridge this gap. As an LLM-powered learning companion, it supports the entire learning cycle, from searching and studying to note-taking and practice, while giving professors the ability to create dynamic, locally hosted knowledge bases that preserve ownership of their course content.
What do our Users need?
“Sometimes I need help just getting started on a topic… it’s not always clear how to approach a problem.”
- Economics Student
.png)
Students
1 Managing Workload
of lab sessions, lectures, assignments, and exams
2 Struggle with Foundational Concepts
Often need additional resources or guidance to fully grasp complex topics
“More staff support is always helpful—not to debug for students, but to teach them how to debug…”
- Head TA
.png)
Teaching Assistants
1 Repetitive Tasks
Often center on key conceptual hurdles like pointers, data types, and fundamental debugging steps
“For answers that TAI gives, it should not be allowed to return code, as this could be problematic if it provides functional solutions directly.”
- Professor
.png)
Professor
1 Onboarding TAs
Dealing with frequent Teaching Assistant turnovers for each semester
2 Multimodal Tools
Current tools struggle to analyze or respond to diagrams and images
Design Challenges
Structuring Complex Knowledge into a coherent, navigable system
Designing AI-supported module learning that is interactive and adaptive
Rethinking Notes as a Reflection Intelligence System
As the design lead, I established 3 key design challenges to tackle. These were drawn directly from our design research and aided in the entire design process.
Student Input
Where is the sweet spot?
System Generated
INTENT
How might we transform static university courses into an interactive, navigable learning system that adapts to students’ mental models while preserving professor-authored structure and intent?
To begin, we knew our solution had to work for every student, including—
The Explorer
Tools that support questioning.
Connections between related topics and concepts.
Difficulty seeing how concepts relate across lectures.
Course structures that feel rigid or restrictive.
The Time Constrained
Quick setup with minimal friction.
Highly prioritized, time-bounded study plans.
Guilt or stress from falling behind schedule.
Wasting time navigating platforms or content.
The Weekly Planner
Clear weekly breakdown of tasks and priorities.
Progress tracking to maintain consistency.
Feeling overwhelmed when multiple assignments cluster together.
Spending time deciding what to study instead of studying.
The Exam Focused
Exam-specific study plans focused on high-impact topics.
Practice questions that reflect actual exam difficulty.
Anxiety caused by lack of feedback before exams.
Static revision guides that don’t adapt to performance.
How did we do it?
Generated customizable study plans
When you first dive into the platform, you’re welcomed with an AI-generated Study Plan built around your course topic. It gives you a clear starting point including key concepts, suggested resources, and practice tasks so you’re never staring at a blank page wondering where to begin.
From there, everything is customizable. Through a context-aware chat, you can remove suggestions that don’t work for you, rearrange learning activities, and ask questions in real time. The plan adapts as you do, pulling in videos, readings, and hands-on practice so you can learn in the way that feels most natural to you.
FOR THE EXAM FOCUSED
To help you actively test your understanding, the platform creates personalized quizzes for each topic, grounded in both your course material and your own notes. These aren’t generic questions but instead reflect what you’ve been studying. After each quiz, you’ll get a clear breakdown of your strengths and gaps, along with smart suggestions to update your study plan so you know exactly what to focus on next.
Adaptive Quizzing and Reflective Feedback
FOR THE EXPLORER
Course Navigation
At the course level, the Topics Dashboard brings everything together. Lectures, notes, videos, quizzes, study plans, and chat history all live in one place, making it easy to jump between topics without losing context.
FOR THE WEEKLY PLANNER
Planning Tools and Temporal Context
As deadlines approach, the system shifts into planning mode. The Exam Schedule shows what’s coming up and instantly generates study plans and quizzes tailored to a specific exam or project. Meanwhile, the Weekly Planner helps you zoom out, reviewing assignments, tasks, and deadlines for the week and turning them into a realistic, structured study plan.
FOR THE TIME CONSTRAINED
And when you need even more control, Tutor Mode lets you take the lead. Select exactly what you want to study, set your time constraints, and generate a custom plan that fits your goals, your schedule, and your learning style.
The result is a study experience that feels less like following instructions and more like having a responsive, collaborative learning partner that adapts to you as you grow.
Tutor Mode
But, how did we get here?
What most AI learning tools lack
Most AI learning tools focus on generic tutoring, broad knowledge retrieval or static content delivery. They offer valuable assistance but often lack deeper contextual understanding, multimodal teaching capabilities, and adaptability to specific courses. Here are the key takeaways from the competitive analysis:
-
Few existing platforms offer AI that is deeply customized to a specific class or instructor, limiting the quality and precision of feedback students receive.
-
Current tools rarely integrate spoken teaching with emotional awareness, reducing their ability to support learners in a more natural, human-centered way.
COMPETITIVE ANALYSIS
PRIMARY RESEARCH
Why TAI?
Given the goal of understanding how students learn within demanding, fast-paced university courses, I conducted primary research focused on their study habits, tool usage, and interactions with AI. Here’s what I uncovered:
Fragmented Tools to Reactive Learning
Students toggle between Google Calendar, Ed, Gradescope, and other platforms, creating scattered workflows that lead to confusion and last-minute scrambles.
Creating a Design System
The visual identity was designed using light tones and simple typography to evoke a sense of comfort, clarity and stability with every interaction.
Lessons
Working on this project pushed me far beyond the boundaries of my previous academic work. It was the first time I built something with such large-scale impact, paired with real user testing across a wide student audience. Navigating that scope taught me how to collaborate effectively with non-designers including engineers, researchers and teaching staff while adapting my process to tight technical and time constraints. More than anything, it showed me how design evolves when it has to move quickly, align with diverse perspectives and still deliver meaningful value to users.