Providing Context
So how did we get here?
Capstone Project
As a senior in Informatics, I partnered with Microsoft for a 6-month capstone to tackle a real-world design challenge: improving adoption of Z.AI, an internal tool for experimenting with prompt engineering using customer support data and Azure OpenAI.Cross-Functional Collabortation
I worked alongside 1 researcher, 1 PM, and 2 engineers, with sponsorship from a Principal Engineer, Data Science Manager, and UX Designer. Together, we balanced user research, technical feasibility, and business priorities.Z.AI is…
An internal Microsoft tool that lets engineers quickly experiment with prompt engineering using secure customer support data and Azure OpenAI. It empowers teams to test ideas safely and provide world-class support.Z.AI is also…
Challenging to learn, with a steep learning curve and usability issues that slowed early adoption. Many users struggled to complete basic tasks, limiting the tool’s impact despite its potential.Providing Context
Why does Z.AI matter?
Enhance Support Efficiency
Reduce errors and streamline workflows so support teams can resolve issues faster.Empower Support Teams
Equip employees with AI-driven tools that improve accuracy and speed in decision-making.Drive AI Innovation
Strengthen Microsoft’s leadership in AI-powered customer service by making cutting-edge tools usable at scale.
Research to solutions process
Uncovered Pain Points
Through user research and heuristic analysis, we identified where Z.AI’s complexity created barriers for adoption.Defined Key Personas
Synthesized findings into two user archetypes (Creator and Consumer), capturing their motivations and frustrations.Designed Targeted Solutions
Created and iterated on design concepts that addressed usability and onboarding challenges.Validated with Users
Tested prototypes with Z.AI users to refine solutions and measure improvements in usability.Our research revealed two distinct user groups that shaped the design challenge:
→ Creators pushed the limits of Z.AI but struggled with its steep learning curve, often losing time to trial-and-error.
→ Consumers valued efficiency and quick wins but were frustrated when the platform prioritized flexibility over clarity.Why these personas matter
Through surveys (30), usability testing (10 participants), and heuristic evaluation, we uncovered consistent friction points across navigation, workflows, and onboarding. These patterns highlighted where Z.AI’s steep learning curve was holding users backResearch Methods
Opportunities for improvement
We translated these pain points into four design opportunities, ensuring each solution targeted a specific user frustration.
User Flows
Based on our personas and pain points, we mapped user flows to capture how Consumers and Creators would move through Z.AI. These flows helped us uncover friction points early and align on opportunities to simplify interactionsUse Case #1
Goal: Create and share experiments with flexibility and control
Key Friction: Overwhelming configuration screens, unclear naming conventions
Design Focus: Streamlined creation flow with clearer inputs and progressive disclosure
Use Case #2
Goal: Explore experiments and analyze results without technical overhead
Key Friction: Too many setup steps before meaningful outputDesign Focus: Reduce clicks, surface default values, and guide users with contextual prompts
Sketches
I started with quick sketches to translate research insights into possible design directions. These low-fidelity explorations allowed me to rapidly test different layouts for onboarding, experiment creation, and support resources before committing to wireframes.Building on my sketches, I translated key concepts into low-fidelity wireframes. The goal wasn’t visual polish, but speed, to test flows, prioritize features, and validate that our design direction solved real user pain points before investing further.
Lo-Fidelity Wireframes
Homepage: Simplified entry point that makes tools easier to find (addresses navigation issues)
AI Chat: Centralized interaction model with clearer flow (addresses workflow inefficiencies)
Learn Page: Central hub for tutorials and resources (addresses lack of guidance)
Experiment Creation: Streamlined form for setup with advanced options hidden until needed (addresses onboarding difficulty)
Home Page Improvements
Redesign: Reduced sidebar to five tabs, added a global search, and surfaced the three most important CTAs (Explore, Create, Chat). Added a Social Hub for community updates.
Impact: Clearer navigation, less cognitive load, and stronger sense of community.
The old homepage was cluttered with unnecessary tabs and text overload, making it hard for users to find core actions.
Before
After
The experiment creation flow had eight steps and no clear guidance, overwhelming new users.
Create Page Improvements
Redesign: Reduced steps to five with a wizard UI, added guidance text and tutorial videos, and introduced progress tracking.
Impact: Faster, less confusing setup that supports both beginners and advanced users.
Before
After
AI Chat Page Improvements
The original page overloaded users with text and technical terms like “temperature” and “top-p value.”
Redesign: Simplified the layout, moved the chat input to the bottom for immersion, and added tooltips to explain advanced settings.
Impact: More intuitive and approachable chat experience, lowering the barrier to entry.
Before
After
Q&A Page Improvements
The old Q&A page was text-heavy and overwhelming, burying critical guidance in long paragraphs.
Redesign: Transformed it into a new Learn page with clear sections for tutorials, FAQs, community discussions, and office hour recordings. Used visuals and structured layouts instead of walls of text.
Impact: Simplified onboarding, reduced cognitive load, and gave users faster access to help and peer support.
Before
After
Our Impact
From our work redesigning Z.AI, Microsoft's LLM experimentation platform, we streamlined prompt engineering workflows and introduced progressive disclosure to simplify onboarding, reduce friction in experiment setup, and improve support efficiency across engineering teams.
Capstone Showcase
For the Informatics Capstone showcase, we set up a display with a poster, prototype walkthrough video, and a video prototype. I designed the poster layout, filmed the walkthrough, helped create the prototype video, and presented the features I led along with the ideation phase. Our design received positive feedback, and we had great conversations with judges and guests throughout the event.
