Corpus Key: Redesigning the AI course creation tool
[Project Overview]
CorpusKey is an AI-powered course material creation tool designed to help instructors generate textbooks, lesson plans, and presentations effortlessly.
Our goal was to transform a developer-centered, complex interface into an intuitive, user-centered experience that empowers instructors to focus on teaching rather than struggling with tools.
As the UX Designer, I led comprehensive user research with instructors and TAs, redefined complex AI-driven workflows into intuitive, flexible content creation experiences, and collaborated closely with CorpusKey’s leadership and development teams to deliver a solution that empowers educators and drives adoption of AI-assisted course tools.
Role
UX Designer
Industry
Non Profit, B2C SaaS
Time
4 months
Tools
Figma
Platform
Web App
Status
Shipped
Impact
Reduced report-writing time by up to 50%
Improved volunteer retention rate
More time for child-focused activities
[Problem Context]
On average, over 70% of the instructors struggled to generate course outlines independently, with most spending extra hours per course setup due to unclear terminology and complex workflows.
For instructors juggling multiple courses each semester, these friction points led to frustration, reduced trust in AI-generated content, and low adoption rates. We conducted a usability test with instructors and TA's and quickly realised the following:
25%
Volunteer Turnover Rate
One in four volunteers leave within the first year, with administrative burden cited as a leading contributor to burnout and disengagement.
120 hours
Lost in administrative tasks
CASA volunteers spend nearly half of their time on court reports and documentation instead of directly supporting children in foster care.
40%
Reduced child interaction time
Volunteers report spending 40% less time in meaningful, in-person advocacy with children than they would prefer due to documentation demands.
[Solution Summary]
We redesigned CorpusKey’s website by simplifying complex hierarchies into clear, nested outlines, introducing draggable content blocks, clarifying terminology, and adding smart preview features.
Rebuilt the information architecture to streamline content creation flows and improve discoverability.
Refined UX copy to reduce cognitive load and support accessibility across academic disciplines.
Redesigned the interface with stronger visual hierarchy and cleaner navigation, enhancing ease of use for both novice and tech-savvy educators.
Introduced a subscription model and freemium pricing strategy designed to drive user adoption while supporting scalable monetization.
50%
Reduction in report writing time
AI-assisted drafting and smart templates cut paperwork effort in half, giving volunteers more time with children.
25%
Decrease in volunteer turnover
Streamlined workflows help reduce burnout, improving volunteer retention and long-term engagement.
2x
Increase in child interaction time
Volunteers can double their face-to-face time with children by minimizing administrative load.
DashBoard
Report writiing
Note Assist AI
Case Summary
In- app messaging
Calendar
[Research]
To deeply understand instructors’ pain points, we conducted user interviews with professors and teaching assistants across STEM and humanities. We also performed a heuristic evaluation of the existing website and analyzed user behavior from platform analytics.
Key findings revealed that over 70% struggled to create outlines without assistance, terminology felt overly technical, and instructors lacked confidence in AI-generated content due to unclear previews and guidance. These insights highlighted a critical need to simplify workflows, clarify language, and build trust through transparent AI interactions.
[Research]
The users said..
4
CASA volunteers
2
CASA Supervisors
1
Foster care system advisor
[Research]
Based on all the data collected through our research we identified 4 key issues
Time drain
Volunteers spend up to 50% of their time on paperwork, leading to emotional fatigue and less child interaction.
Collaboration Struggles
Coordinating with professionals and families requires frequent, sometimes difficult, communication.
Information Overload
Managing large amounts of data makes it hard to extract key insights quickly.
Emotional disconnect
Volunteers feel their purpose of building relationships with children is overshadowed by admin work.
Legal Anxiety
Fear of missing or misreporting case details adds stress, reducing volunteer confidence.
Fragmented Tools
Volunteers often switch between multiple apps, emails, and printed notes, increasing errors and frustration.
[Analysis]
To translate our research into actionable insights, we mapped the entire CorpusKey user flow and created an affinity diagram to categorize pain points, suggested solutions, and small delights.
[Analysis]
The site map analysis helped us pinpoint critical friction areas, such as unclear entry points, ambiguous back navigation, and confusing prompt flows. This clarified exactly where instructors felt lost and overwhelmed.
Through affinity mapping, we clustered key pain points around lack of feedback, missing hierarchy, unclear instructions, and trust barriers. These insights directly shaped our design priorities and highlighted opportunities to simplify the platform and empower instructors to work confidently without external help.
[Analysis]
From this synthesis, I derived core design principles to guide solution development
Administrative overload
Volunteers felt “administrative overload” overshadowed their core mission.
Lack of trust in digital tools
There was a lack of trust in digital tools, causing continued reliance on paper notes.
To give emotional support
Volunteers desired more emotional connection and confidence rather than just efficiency.
Simplifying the process
Simplifying compliance was as important as simplifying UI.
[Analysis]
From this synthesis, I derived core design principles to guide solution development
Time over tasks
Prioritize solutions that give time back rather than simply digitize existing workflows.
Confidence through clarity
Ensure volunteers feel secure and guided, especially with legal documentation.
One central source
Minimize tool fragmentation and create one trusted system.
Inclusive by default
Design for accessibility as a baseline, not an afterthought.
[Ideation]
Guided by core design principles, we explored multiple ways to simplify workflows, clarify structure, and build trust in AI-generated content.
We started with low-fidelity sketches and wireframes to quickly test ideas such as nested outline structures, draggable content blocks, and contextual helper text. We also experimented with guided walkthrough flows and progressive onboarding to support independent exploration.
[Ideation]
During this phase, we prioritized rapid iteration and early feedback from instructors, allowing us to validate mental models and uncover usability issues before moving to high-fidelity designs.
This approach helped us align our solutions with real instructor needs while keeping the business goals of adoption and retention front and center.
[Ideation]
The ideation process revealed that we would focus on the following aspects to improve the overall design
A web-based platform not only allowed us to integrate AI-powered features seamlessly but also ensured ease of access across devices. This decision laid the foundation for the overall information architecture and interaction flows, summarized in the initial site flow and site map below.
[Iterations]
To quickly test information architecture, workflows, and key interaction patterns, we began with low-fidelity wireframes and flow sketches.
These early explorations allowed us to validate fundamental concepts with CASA volunteers and stakeholders before investing in detailed UI design.
[Testing]
Participants were asked to create outlines and reorganize modules using drag-and-drop. Feedback revealed improved comprehension of workflows, higher trust in previews, and reduced reliance on external support
Key metrics included a 40% reduction in outline setup time, a significant increase in System Usability Scale (SUS) score from 55 to 85, and stronger positive sentiment around the simplified language and guidance.
Iterative refinements based on this feedback helped us align even more closely with instructors’ mental models and usability expectations, ensuring the final experience felt intuitive and empowering.
Navigation and wayfinding
Some volunteers struggled to understand the sidebar structure and where to find specific case-related reminders and files.
AI note assist trust gaps
Volunteers expressed initial hesitancy in relying fully on AI-generated content, fearing it might miss nuanced child information or legal phrasing.
Onboarding overwhelm
Older volunteers mentioned needing a simplified first walkthrough to feel confident using the system independently.
Reduced cognitive load
Volunteers completed draft reports 40% faster, feeling less overwhelmed by paperwork.
Centralized dashboard clarity
Participants appreciated having all child case details, upcoming tasks, and notes in one place.
Positive emotional response
Volunteers felt less anxious about forgetting important details or making legal mistakes
[Impact]
Our redesigned CorpusKey experience empowered instructors to create course materials more confidently and efficiently, while also aligning with business growth goals
Beyond metrics, users reported feeling more in control of their content and expressed higher trust in AI-generated outputs — a crucial factor for driving long-term adoption and brand loyalty.
50%
Reduction in report writing time
AI-assisted drafting and smart templates cut paperwork effort in half, giving volunteers more time with children.
25%
Decrease in volunteer turnover
Streamlined workflows help reduce burnout, improving volunteer retention and long-term engagement.
2x
Increase in child interaction time
Volunteers can double their face-to-face time with children by minimizing administrative load.
[Reflection]
As a UX Designer,
This project challenged me to balance deep user empathy with business strategy, and to redesign a highly technical tool into an approachable, educator-first experience.
What I learned:
The importance of aligning AI-powered workflows to mental models, rather than forcing users to adapt to system logic.
How strategic microcopy and transparent previews can build trust in complex, unfamiliar technologies.
The value of early, iterative testing to uncover usability issues and validate design direction before high-fidelity investments.