December 2024 - May 2025
The HCDE capstone project is a collaboration with Microsoft Azure Communication Services team
Designers and researchers from ACS team, 3 students
Overview
Problems
Fragmented Communication Across Platforms.
The rise of digital tools like E-mail, Slack, Canvas, and Zoom has made managing communication across them challenging for both teachers and students.
Solutions
AI assistant on Canvas (LMS) to Streamline Communication Between Educators and Students.
Results
Improved efficiency and clarity by reducing platform switching, ultimately saving time.
Student Kim
“…at least in this interface, it seems cleaner than all the messy messages in the inbox.”
Educator Jessica
“You have created something that would be a time saver, because the aggregation of student questions is quite useful.”
Video
Click to view our product video!
Research
Research Questions
How can AI help streamline and support meaningful interaction between teachers and students across multiple platforms?
We conducted user interviews, literature reviews, and participatory workshop to gather user insights.
Research Findings
Similarities between students and educators
Finding #1
Multi-platform use
Students and educators use various tools (email, Canvas, Slack), but students bear the burden of adapting to each.
Finding #2
Missed messages
Both groups struggle with notification overload, leading to missed important messages, especially during busy periods.
Finding #3
Desire for feedback
Educators want to know if messages are seen, while students seek quicker responses to reduce anxiety.
Different behaviors
Finding #4 Boundaries
Finding #5 Attitude towards AI
Students are generally more open to using AI in education while some educators have some hesitation towards AI.
Concept
How can we address user pain points and meet their needs?
Minimize platform switching to simplify communication.
Design an AI assistant integrated in Canvas (LMS) using ACS components to streamline communication.
By integrating this AI assistant, users can manage all their communication in one place. This helps centralize conversations, keep everything organized, and reduce notification overload.
And the AI assistant can access course data and user activity, which can provide timely, context-aware responses to support better communication.
Design Challenge
Ideation
Brainstorm
For each user painpoints, we brainstormed several solutions and then grouped them into three features.
Customization
Smart prioritization
Reminders of important tasks
Quick reaction to message
Automation
Course content creations
Automate tedious tasks
Schedule meetings
Contextual AI Support
AI-generated answer
Help student ask questions
Act as a thought partner
Jobs to be Done
For Student
For Educator
Wireframes
I translated these ideas into wireframes, which helped us visualize key interactions and quickly iterate.
E.g. How to integrate Lumo into the chat to enable context-aware support for users?
Side by side
Clear human vs. AI distinction.
Risk of losing context and fragmentation.
✅ Embed into the chat (Selected)
Better context continuity and easier follow-ups.
Confusion about who is chatting with.
We chose the second option to embed Lumo directly into the main chat and added visual indicators to clearly show whether the user is speaking with a human or the AI.
Prototypes
I created a prototype to illustrate how Lumo could be embedded into the LMS and to demonstrate the user flow.
Testing
User Testing
We conducted user testing to validate our design decisions, understand user behaviors, and identify usability issues.
We tested with students and educators about the entry points, dashboard, messaging, interactions for switching between human and AI, and the announcement feature.
Findings and Iteartion
Finding #1 Dashboard and insights need clearer purposes
Both students and educators were confused by AI Insights and the "What's on your mind" section.
Improvement #1 Prioritize dashboard content based on importance and reword key sections to enhance clarity for users.
Finding #2 AI messing is helpful but needs customization
Educators valued AI-generated summaries for saving time, especially for repetitive questions. Students liked AI phrasing suggestions but wanted control to maintain their own tone.
Improvement #2 Emphasize that AI messaging is optional, and provide clearer recipients between AI and human.
Finding #3 Feedback mechanism was unclear
Users were unsure how to use the thumbs-up/down on insights and how to find the previous ones.
Improvement #3 Clarify the purpose of feedback buttons and add a new tab to separate chat and insight history.
Finding #4 Users Need Context Behind AI-Generated Insights
Both students and educators wanted to understand the reasoning behind AI-generated insights. They looked for supporting context such as relevant resources, or direct links to the original content.
Improvement #4 Provide clear explanations, source references, or embedded links to help users trust and act.
Conclusion
Deliverables
Instant Q&A Support
Lumo helps students get quick, accurate answers to their questions.
Smart Message Drafting
Assists students in crafting clear and effective messages for better communication.
Automated Response Assistant
Enables educators to efficiently respond to frequently asked questions.
Insights & Announcements
Analyzes recurring questions and suggests relevant actions as posting announcements.
Responsive Design
Impact
Helped users reduce repetitive replies, improve communication efficiency, saved hours of weekly effort.
Student Emily
" It's nice that you don't have to give—like, I used to just copy a ton of text and ask, ‘what's going on’ to my other AI platform. So it's just nice that you don't have to copy everything there."
Educator Allen
“It definitely helps me avoid that repetitive replying. If multiple people have asked the same thing, it allows me to understand that without having to go through all of them one by one…”
Design System
Limitations
AI in Education
When implementing AI chatbots in education, it’s important to proceed with caution due to strict policies and regulations. These rules are in place to protect student privacy and data security, so any use of AI must be transparent, ethical, and compliant with institutional standards.
Different perspectives about AI
