I integrated Generative AI to provide personalized guidance and learning suggestions to 200,000+ global student developers who feel lost on Microsoft Azure, helping them navigate their learning journey with cloud computing better.
Timeline
June - September 2024
(3 months)
Role
Product Design Intern
UX Lead
Collaboration
Product Manager
Content Designer
Engineers
UX Designers & Researchers
Impact
Users find the learning content they want 60-65% faster
100% user satisfaction rating
CONTEXT
What is Azure for Students (A4S)?
This is an educational platform on Microsoft Azure that was created specifically for student users to…
develop cloud computing skills through courses and learning paths
get familiar with Azure services and grow into long-time users
PROBLEM
Students get lost and don't know how to get started
I conducted a survey with 130+ college students in Computer Science and related majors across the U.S. to understand their backgrounds and challenges.
Analyzing Net Promoter Score (NPS) and comments from 1000+ users on Azure for Students, I realized the platform was providing generic learning recommendations instead of listening to individual needs.
What Azure provides
"Popular courses to take"
What students want
"Study plan curated for you"
How might we provide learning direction
that caters to each student and their unique needs,
as well as improve their engagement with the platform?
Exploring how AI can be a great tool for customizing user experience on other educational platforms (e.g., Coursera, LinkedIn Learning, Duolingo), I decided to adopt the idea, but with an additional feature.
I worked with a content designer to decide on which questions to ask new users to help them build a study plan. The quiz should collect the most helpful information while not exhausting users out with so many questions.
After finishing the quiz, users will be provided with a personalized study plan guiding them through 3 stages of the learning journey.
Copilot not only helps first-time users build a study plan, but also serves as a long-term companion that provides suggestions to update the plan reflecting learners’ changing interests and progress.
My initial idea was to integrate inline AI suggestions into each piece of learning content.
Iteration #1
❌ Shifts the layout of the whole page and feels glitchy
Iteration #2
❌ Icon is small and may be overlooked
Gathering feedback from users and stakeholders, I realized "Copilot Suggestion" is an important feature that needs its owns system, not just a pop-up that can be overlooked.
Therefore, I incorporated the feature into the AI chatbot on the side panel, enabling users to interact with the AI to refine the plan based on their needs.
Although the new design and features haven't been implemented or shipped yet, the usability testing sessions I conducted with users indicate that we can expect significant impact from the improvements.
~ 200,000 students
around the world will benefit from the design improvements.
2:14 mins ⇒ 35s
Users will spend less time navigating and figuring out the new Overview page than the original page.
100% students
who tested the new design (6 out of 6 users) expressed excitement about the new Customized Study Plan feature and indicated that they would use it once it’s launched.
⇑ 5-10 NPS points
Net Promoter Score (NPS) of the product is expected to increase, as the user satisfaction on the platform is also expected to be improved.
Balancing user goals and business goals
Although this was the most challenging part of my project, I got to learn to communicate effectively with cross-functional stakeholders and users, and to go the extra mile by conducting additional research and usability testing to support my design.
Leverage AI to solve user problems
I had a great time (and also the first time!) designing an AI-powered feature for my product, especially when I decided to think outside the box and explore ways to implement Copilot beyond just a chatbot.
Design is an evolving process
Juggling multiple user/business requirements, I felt the pressure to finalize a design that could satisfy everyone. However, design is an iterative process, and there’s not really a “final” design. What I delivered was the best possible within the internship's time constraints, and there'll always be room for improvement.
So, keep learning and moving forward!
Beside the internship, I've had a wonderful summer exploring Seattle, WA, making new friends, and going on hikes! Truly grateful for the internship experience and the memories I had with my amazing people.
See you again, Seattle!
