Microsoft Copilot
Designing learning inside AI — turning curiosity into confidence
Internship Project
AI
UX
User Research
Team
Microsoft Copilot Core UX
Duration
May 2024 - August 2024
Role
Product Design Intern
Skills
Cross Collaboration, Design Systems, Interaction Design
OVERVIEW
What did I do?
During my 12-week internship at Microsoft as a UX Design Intern on the M365 Core UX team, I focused on enhancing learning and first run experiences within Microsoft Copilot. My main project, I worked with a Product Manager Intern and an Engineering Intern as a triad to create a Copilot Help Extension, aimed to support users’ learning and discovery directly inside Copilot — helping them move beyond trial-and-error exploration and improving retention.
PORJECT SCHEDULE
Internship Timeline
During the first few weeks of my internship, I explored ideas around Regenerate and Refinement in Copilot. However, as team priorities shifted, my project pivoted toward a new problem space. enhancing learning and discovery within Copilot.
DISCOVER
What is the problem?
According to recent user research, learning is currently not supported inside of Copilot. Customers are looking for help that goes beyond just the suggested queries we show today, which may be a factor in our new user retention being a critical issue.
DISCOVER
How do new users learn how to use Copilot right now?
Insight: Customers having to rely solely on experimentation may lengthen the learning curve, which can lead to frustration and hinder Copilot adoption
DISCOVER
From the user
There’s a lot of learn. It’s like a car - it moves, great. But learning how to drive it is another thing. [I need to learn] how to drive the car that is Copilot
I was hoping to see tutorials - a quick intro to Copilot
Insight: Users in the study express a need to get help learning how to use Copilot. With respect to FRE, current in product learning mechanisms are not doing a good enough job at supporting learning
DISCOVER
Why is this a problem - Learning Effects User Retention
In the real world, we would lose user here
By the end of the first week, users’ initial excitement about Copilot began to fade as they encountered its limitations. Outside of a research setting, this decline in sentiment would lead to early drop-off. One of the key drivers behind this loss of confidence was that learning within the interface wasn’t well supported, leaving users unsure of how to continue engaging with Copilot.
DISCOVER
Current Experience
No First Run Experience (FRE), users get dropped into the Query page
Lack of Guidence
No Structured Onboarding
Unhelpful response when asking Copilot help question
Vague
Incomplete
Not Actionable
PROBLEM SPACE
The lack of a learning experience in Copilot creates a disconnect for new users in discovering essential knowledge and functionalities
PRODUCT ASK
Create a Copilot Help Extension that offers customers a dedicated guided chat experience focused on the specific learning problems identified in user research
DISCOVER
MVP Project Constraints
Build on GPT/Extensions framework
Triad Intern Project
Relatively Easy and Quick to Build
IDEATE
Objective
Defining Entrypoint to Copilot Help
Discoverability
1
Experience within Copilot Help
Help Extension
2
Learning Pillars
3
Larger Learning pillar to help users onboard to Copilot faster
Project Objectives
Click to skip objectives!
Discoverability
Identifying key moments in the user journey where learning support should surface—so help feels contextual, not disruptive.
1
OBJECTIVE ONE
IDEATE
What are some good opportunity to introduce Copilot Help?
When users first open Copilot, they’re eager to understand what it can do—but without structured guidance, this curiosity often turns into uncertainty. Providing learning support early helps users build confidence from the very first interaction.
During Initial Use (First Run)
Opportunity 1
When users ask “how-to” questions or receive unsatisfactory responses — moments that reveal curiosity but also confusion. These instances are key opportunities to surface guidance and help users stay engaged instead of feeling stuck.
Within Conversation
Opportunity 2
Entrypoint #1
✓ Anticipates user needs
✓ Proactive Assistance
✓ Provide prompt suggestion
✓ Direct introduction
Embedded directly into Copilot’s chat interface during the first five uses, these cards feature distinct visuals callout and a short message that prompts users to enter the extension.
Proactive Tip Cards
Copilot Help Extension
Intern Final Presentation
Entrypoint #2
✓ Streamlined interaction
✓ Just in time help
✓ Part of a response to relevant queries
When users ask “how-to” questions or receive unclear responses, a subtle message appears within the chat. These context-aware guiding users toward Copilot Help without breaking the conversational flow.
Conversational Nudge
Copilot Help Extension
Intern Final Presentation
IDEATE
Additional Design Explorations
Conversational FRE
Fluent Card FRE
Feedback
Feedback
Natural introduction to Copilot
Lightweight implementation
Emphasizes conversational AI and multi-turn dialogue
High discoverability
Parallel projects underway
Limited learning depth
Entry points should be more integrated and less disruptive
May be overlooked or dismissed quickly by experienced users.
An onboarding chat that introduces users to Copilot through an AI-led conversation.
Lightweight contextual cards introducing Copilot Help during the first few uses.
Defining Experiences Within Copilot Help
The next step was to define how users experience learning once they enter Copilot Help Extension.
2
OBJECTIVE TWO
COLLABORATION
Close Collaboration Across Team
UX Intern (Me!)
Engineering Intern
PM Intern
Content Designer
User Researcher
Bringing the Copilot Help Extension to life required tight collaboration across multiple roles — from product management and engineering to content design.
Feature Priority Alignment
Worked closely with PM, engineering and research partners through daily syncs to prioritize features, scope dependencies, and ensure a feasible, high-impact MVP that could be built and tested internally.
1.
Design–Engineering Integration
Partnered with engineers through build reviews and bug bashes to refine interactions, polish UI details, and maintain design fidelity across implementations.
2.
Content Collaboration
Collaborated with the content designer to fine-tune the language of help responses — from phrasing and tone to instructional clarity — ensuring guidance felt approachable, consistent, and human.
3.
SOLUTION
Copilot Help Extension
Intern Final Presentation
By the end of my internship, we launched a working Copilot Help Extension, enabling internal users to ask questions about how to use Copilot and discover how it could benefit their daily work — all directly within the Copilot interface.
✓ Internal-ready build
✓ Provide accurate and helpful response
Copilot Help Extension Live Demo - It’s in Build!
Future Visioning: Copilot Learning Pillars
With the initial build completed, our focus shifted toward the future of learning inside Copilot
3
OBJECTIVE THREE
FUTURE VISIONING
Aligning Vision and Impact Prioritization
To shape the long-term direction of learning within Copilot, our team began by hosting a brainstorming session in FigJam with key stakeholders, including designers, PMs, and researchers.
✓ Central Learning Hub
✓ Learning Resources
✓ Personalization & Job Role
✓ Social Sharing
✓ Progressive Disclosure
After capturing a wide range of possibilities, we plotted each concept on an Impact–Effort Scale to evaluate feasibility and potential user value
INSIGHTS
Key insights from brainstorming session
Central Learning Hub
Learning Resources (ie. Video, Tutorials)
Personalization & Job Roles
Social Sharing
Progressive disclosure
From these, a unifying concept emerged: the Copilot Learning Hub — a centralized ecosystem that brings all learning pathways together.
FUTURE VISIONING
Copilot Learning — Longterm Vision
Design Principle: Integrated learning and support that meet customers where they are
Learning Hub
A central place within the Copilot interface where users can find all help-related resources.
Contextual Help Scenarios
Context-aware assistance directly within the Copilot. When users ask for help, Copilot responds with relevant answers + additional resources.
Before diving into solutions, we stepped back to define what “learning inside AI” should look and feel like, creating design principles that grounded our decisions in clarity, empathy, and purpose. These principles ultimately shaped two key solutions that bring the idea of integrated learning to life: the Learning Hub and Contextual Help.
SOLUTION 1
Learning Hub
Copilot Help Extension
Intern Final Presentation
✓ Different Learning Preferences
✓ One-stop destination
✓ Cohesive Support Syste
The Learning Hub was envisioned as a one-stop destination within Copilot where users can explore tutorials, videos, and guides — all without leaving their workflow. By centralizing guidance in one place, the Learning Hub turns scattered information into a cohesive support system inside Copilot.
SOLUTION 2
Contextual Help Scenarios
Copilot Help Extension
Intern Final Presentation
✓ Resource Linking
✓ Integrated Learning and Support
✓ Intent Recognition
Building on the foundations of the Copilot Help Extension, Contextual Help reimagines learning as something fully embedded within the Copilot interface. This design enables immediate, context-aware learning.
REFLECTION
Here is what I learned during the process...
At first, it was challenging to dive into a product area that hadn’t been fully explored before. I knew I had to step out of my comfort zone, so I proactively reached out to cross-functional partners both within and outside of my team to gather insights and feedback. Through this process, I was able to build a clearer understanding of the landscape, approach problems from different perspectives, and create innovative solutions that truly made an impact.
💡 Stay Curious
It’s not just about finding solutions—sometimes, asking the right questions and reflecting on the journey leads to the best outcomes.
🔍 Embrace the Process
Sometimes, the best insights come when you pause, reflect, and reassess before moving forward.
🚀 Take a step back and reflect
RELFECTION
Internship Highlights
Cross-functional Collaboration
Consolidating Feedback & Developing POV
Creating Designs Base on Dependencies and Timlines
RELFECTIONS
My internship was an incredible journey filled with learning, creativity, and connection. Over the 12 weeks, I had the privilege of working with some of the most talented and kind-hearted people. It wasn’t always smooth sailing—there were plenty of ups and downs along the way. But through their guidance and collaboration, I not only expanded my skill set but also grew significantly as a designer. This summer was full of bold ideas, meaningful growth, and invaluable lessons, and I’m grateful for every moment and everything I was able to learn!
My main takeaway
OUTSIDE OF WORK
I Got to Enjoy the Beautiful Summer in the PNW ☀️
Also check out…↓
FedEx Internship
Redefining the print customization journey for both customers and employees
MindSpace
From motivation to maintenance: making self-care a habit