navigate

navigate

navigate

navigate

Chat interface with Topper, an AI chatbot, offering course selection help on a clean, light blue background.
Chat interface with Topper, an AI chatbot, offering course selection help on a clean, light blue background.
Chat interface with Topper, an AI chatbot, offering course selection help on a clean, light blue background.

Topper Advisor

An AI chatbot that streamlines elective selection and tailors degree pathways for university students, easing the advising workload.

Client

St. Edward's University

Season

Spring 2024

Timeline

10 Weeks

project overview

project summary

Topper Advisor was developed to simplify the elective selection process for students at St. Edward's University. Utilizing a conversational AI, the chatbot aids students in finding courses that match their schedules and academic goals by providing detailed course information. The project aimed to improve user satisfaction and reduce the workload of academic advisors.

team + role

Myself and three other students collaborated in ideation, research, design, prototyping and development of this product. During the development phase myself and one other student developed and trained the chatbot using OpenAI Assistants and received guidance in connecting the API with Flask from an external software developer while the other two students enhanced our database in google sheets filling in additional information about courses from the catalog and additional sources.

challenges

Identifying Gaps: Currently, St. Edward's University lacks a systematic approach for students to explore and select elective courses that complement their major and minor studies, resulting in potential gaps in both academic and personal growth.

goals

Enhanced Discovery & Autonomy: Our primary goal is to empower students to independently discover and select electives aligned with their personal interests and career goals, while enabling academic advisors to manage their advising workloads more efficiently.

outcomes

Streamlined Processes & Personalized Education: Implementing this solution will streamline elective discovery, boost student satisfaction by offering tailored course recommendations, and reduce the administrative load on academic advisors.

discovery highlights

Lean UX Canvas

We initiated our research with a Lean UX Canvas workshop, collaborating across the class to identify the most pressing issues within the existing advising system. This canvas helped map out the main pain points, risks, goals, and opportunities, including the growing workload of academic advisors and the limited autonomy students have when selecting electives.

Proto Personas

Through stakeholder analysis, we identified two key direct user groups: academic advisors seeking workload reduction and students needing personalized academic planning. We developed detailed proto personas representing diverse student scenarios—transfer students, commuters, first-generation students, and those balancing complex schedules—ensuring our solution could accommodate varied needs and backgrounds. This comprehensive approach to understanding our stakeholders provided clarity on specific user needs and challenges, giving us a solid foundation for ideation.

current Landscape

After identifying the problem areas our next step was analyzing the university's existing course scheduling system. The current tool presents significant limitations that impact both students and advisors. While it allows basic searches by subject, course number, and keywords, it lacks critical functionality for effective schedule planning.

identified limitations


  • No ability to filter courses by time of day

  • No option to search by specific days of the week

  • Limited filtering capabilities for course modality (online vs. in-person)

  • Basic search interface that requires students to manually cross-reference course times

  • No integration of course prerequisites or degree requirements

  • Absence of personalized recommendations based on student interests or degree path

limitation impact

Often force students to rely heavily on advisor support for course planning, as they must manually review each course's schedule details to determine compatibility with their existing commitments. This creates additional workload for advisors who must assist with what could be an automated process.

initial survey

We conducted a survey among St. Edward's students to gain a better understanding of what mattered most to them and their experience in selecting electives. This insights below highlight our survey insights.

77.7%

faced challenges finding and comparing electives

70%

prioritized course topics in their selection

90%

had strong preferences for course modality

66.6%

struggled to find electives matching their interests

process highlights

resources + landscape

technical tools

Evaluation of Technical Tools

The team identified and evaluated five AI chatbot platforms, including Bot Penguin, Landbot, Botpress, WatsonX, and custom GPTs, to determine the best technical tools for their AI Advising Assistant

Foundation with Figma Prototype

Before proceeding to more complex development steps like API development, dataset creation, and model training, the team decided to establish a solid foundation by first creating a Figma Prototype

Focused Assessment of Benefits

This foundational approach was guided by a focused method of using a subset of 32 elective courses to assess the potential benefits an AI chatbot could offer compared to traditional elective discovery methods

market landscape

Diverse Research Methodologies

The team conducted research that included a survey administered to students to validate assumptions and secondary research into chatbots and user perspectives

Investigation of API Integration Tools

As part of their process, the team also performed research related to API integration tools

User Flow Design

The team drafted user flows to define the logic for the system, basing these on their initial sample data

identified risks and conditions

identified risks


  • What if the chatbot produces misinformation about courses?

  • What if students don’t trust the chatbot?

  • What if we don’t understand the questions students have about elective courses?

  • What is the right amount of questions it can ask the students before they become frustrated or overwhelmed?

risk + hypothesis tests


  • Narrow in on what features the chatbot could provide

  • Conduct research about student’s current experiences and questions about advising

  • Conduct secondary research about AI Chatbots and user perspectives.

acceptance conditions


If the chatbot effectively understands and processes student's inquiries and preferences, it will boost academic advising efficiency and help students independently find electives personalized for their degree pathway.

flowcharts

We mapped Topper's conversation flow to mirror real academic advising:


  • Starts with student interest areas

  • Presents relevant elective options

  • Explores course details on request

  • Provides scheduling information


This flowchart guided our prototype development and ensured natural conversation paths.

design + branding

Bot Penguin's clean interface influenced our minimalist design approach, while St. Edward's brand colors ensured visual consistency. For the bot's personality, we created a scholarly interpretation of our mascot—a professionally-dressed Topper the goat—distinct from the athletic branding to reflect the academic nature of the tool.

Bot Penguin's clean interface influenced our minimalist design approach, while St. Edward's brand colors ensured visual consistency. For the bot's personality, we created a scholarly interpretation of our mascot—a professionally-dressed Topper the goat—distinct from the athletic branding to reflect the academic nature of the tool.

prototyping

We transformed the initial flowchart into a dynamic Figma wireframe, incorporating real course data and interactive elements. This fully interactive prototype allowed us to conduct usability testing, gathering critical feedback on Topper’s functionality and user engagement, aligning our developments closely with student needs.

user testing

We conducted usability tests with six participants, guiding them through simulated course selection tasks using Topper's prototype. The feedback helped validate our design decisions and identify areas for improvement.

insights

  • Step-by-step enrollment guidance was highly valued

  • Detailed course descriptions received positive feedback

  • Underclassmen found more value than upperclassmen

  • Personal response style increased engagement

Key Pain Points

  • Insufficient course details (room numbers, credits, prerequisites)

  • Unclear focus area groupings

  • Generic responses lacking personalization

  • Limited appeal for experienced students

Design adjustments

  • Added comprehensive course information

  • Restructured focus areas for clarity adding key words in our database

  • Implemented personalized response patterns through model training

  • Added career-aligned recommendations

stretching into development

Moving from prototype to product, we transformed Topper into a dynamic advising tool using OpenAI's Assistant API and Flask. This shift enabled more natural conversations and personalized recommendations, significantly enhancing the chatbot's value for students.

technical foundation

  • Built with Flask + OpenAI Assistants API

  • One-week rapid development timeline

  • Enhanced with comprehensive course database

  • Streamlined interface from Figma design

key improvements

  • Natural language processing for flexible queries

  • Real-time course information access

  • Personalized recommendations based on interests

  • Dynamic response system vs. predetermined paths

team collaboration

While developers focused on API integration and functionality, other team members enriched our database with detailed course information, creating a more robust advising tool.

feature highlights

Personalized Elective discovery

challenges addressed

Traditional course selection tools offer limited personalization, requiring students to manually search and compare options. This leads to overlooked opportunities, time-consuming research, and heavy reliance on advisor support for basic course discovery and validation.

solution overview

Topper combines AI-driven personalization with interactive learning to transform elective selection. The system analyzes student preferences, learning styles, and career goals while enabling natural Q&A exploration of courses. Through continuous feedback and pattern analysis, it surfaces relevant options that might otherwise be overlooked.

hypotheses projection

By providing intelligent course discovery, we expect to reduce advisor workload while helping students make more informed decisions about their electives. This personalized approach should lead to better academic outcomes, increased student satisfaction, and more efficient use of advising resources.

status + reflection

project status

This project was part of a senior capstone project and has concluded. All files and insights have been passed on to relevant departments at St. Edward's University and is undergoing further review and analysis by current students in the User Experience Program to evaluate possible implementation.

reflection

In this project, I ventured beyond my comfort zone, harnessing multi-disciplinary skills in UX and AI to transform a standard course catalog into a dynamic, conversational interface. This not only met a practical need but also fused student needs with institutional goals, illustrating the potential of technology to foster empathetic, personalized user experiences.