Context
In traditional university courses, learning is assumed to flow linearly:
Learn → Practice → Evaluate → Exams
This is where TAI comes in:
a course-integrated, fine-tuned teaching agent built to support deep learning, not just quick answers.
But real learning rarely works that way. Many students get stuck in the concept phase, without timely help.
“I’ll just ask ChatGPT… it’s fast.”
But fast answers don’t lead to deep learning.
In the LLM era, a new gap is growing:
✅ Task completion
❌ Knowledge mastery
Design Overview
Check it out
Biggest Challenge
Working with LLMs in education revealed complex tensions:
Control vs Openness
Clarity vs Overload
Trust vs Automation
How to keep answers aligned with course scope
How much to explain without overwhelming
Professors need oversight without micromanaging
How might we support students during moments of confusion with course-aware, personalized AI guidance—without increasing instructor workload?
Design Process
Due to the novelty of LLMs in education, we conducted interviews with:
5 professors (CS, Cognitive Science, Education)
4 students (CS, Econ, Data Science)
Design Thinking: From Dual Needs to Scalable System
User Flowchart
What Makes Us Different
Secure, Specialized, and Purpose-Built Model:
trained with clear ethical guardrails
task-specific AI agents keep students focused on course objectives
each course with its own model that has defined curriculum scope
Ethical check
Out of scope check
Alignment with Course Scope & Progress
responses always refer to official course materials, ensuring correct learning boundaries and concept depth
understands course structure and prerequisites, introducing new material based on what students have already mastered
adapts to different instructional styles, seamlessly blending theoretical knowledge with practical exercises
✨ Reflection & Next Step
Next Step
Refine UI/UX with sharper micro-interactions
Broaden user testing for diverse contexts
Explore integration with educational platforms to scale impact.
Reflection
This project deepened my understanding of how design and technology push each other forward. I learned the value of rapid prototyping, clear communication, and crafting AI interactions that feel intuitive and approachable.