Accelerate your course learning with tailored Study GPTs
Context
What's Coursepals.AI?
Coursepals.AI is an all-in-one personalized GPT application for course learning and academic research. With Coursepals.AI, you can enhance your study routine by completing assignments up to five times faster using tailored GPTs for your courses while enjoying built-in access to academic sourcing APIs.
But what does 1.0 version look like?
Business Requirement
Resolving Issues Uncovered During the 1.0 Demo Launch
While promoting Coursepals.AI at Stanford, we encountered challenges and realized the B2B approach was unsuitable for the product. We faced these two main marketing challenges:
After assessing our funding and resources, we realized we couldn't sustain such a long period without revenue. We decided to pivot from B2B to B2C, shifting our target users from professors to students.
Research & Interviews
Contrasting Needs of Students and Professors
We conducted 12+ interviews and countless surveys in Stanford Confessions to learn more about their pain point and willingness to purchase. We then developed the persona and user journey of our new target users.
Challenge
Users struggle with understanding and motivation to use personalized GPTs
Based on users' lack of understanding in AI chatbot creation and AI product terms, I identified two core design tasks: simplifying the process and clarifying terminology to boost motivation and usage.
How I solve task 1
Started by GenAI product design thinking
I adopted GenAI product thinking to guide the UX design and use flow. Two key needs drove our approach: 1) streamline the user workflow and 2) reduce the learning curve while improving efficiency.
Leveraging fundamental GenAI design principles— automation of repetitive tasks and minimal user input —we transformed the new use flow of creating instructions.
Obstacles Identified in Testing
Initial solution was stalled by both users and the dev team
We quickly got some feedback from the dev team and our beta users, although not positive feedback.
Iteration
Refocusing on core user needs over direct GenAI solutions
I was curious about why we received so much seemingly negative feedback, so I decided to communicate with both users and the dev team to gain a deeper understanding.
With the users, I explained what 'instruction' means and guided them through our onboarding process.
With the dev team, I learned how the specific tasks the back end was handling during the instruction generation process.
Final Solution
Prioritizing instruction generation, reducing analyzed files
Coursepals.AI focuses on specific educational use cases, setting it apart from ChatGPT. While ChatGPT enables users to create their own GPTs, it’s still hindered by file upload limitations and a steep learning curve in setting up workflows. This complexity has led many users to overlook or avoid using the ‘Create Your Own GPT’ feature, especially when it comes to assisting with academic tasks or essays.
Impact
Successfully Improved GPT Creation Completion Rate and Efficiency
I achieved a significant impact with the new solution, leading to notable improvements in both the completion rate and efficiency of GPT creation.
How I solve task 2
Classic approach to familiarize users with new terms
APIs play a crucial role in personalized GPTs, but many of our target users are unfamiliar with them. By enabling users to search and generate answers from various academic knowledge bases through different APIs, we recognized the need for a smoother introduction.
To address this, we implemented 2 proven strategies within Coursepals.AI to help users seamlessly familiarize themselves with APIs.
Revision
Design APIs-switching feature for intuitive user exploration
The solution is effective, and beta users seem satisfied, but I felt it lacked intuition. I asked myself, 'What's the best way to teach users?' With this in mind, I designed an API-switching feature to guide users naturally without relying on complex explanations. I then also conducted A/B testing with beta users to see their reactions to the revisions.
Takeaway
Designing AI product: enhancing user experience without sacrificing engagement
Designing AI products doesn’t mean relying on AI to replace user effort and assuming it leads to a better experience.
Early in the design process, especially as it was my first time creating a consumer-facing AI product, I found myself focused on showcasing the intelligence of the AI.
However, I realized that relying solely on AI to solve user problems can be a shortcut. It’s essential to strike a balance between leveraging AI’s capabilities and thoughtful design, ensuring users stay empowered and engaged, rather than feeling overwhelmed or disconnected.
