AI ADVISOR

A mockup of AI advisor onboarding on a computer.
A mockup of AI advisor onboarding on a computer.
A mockup of AI advisor schedule editing on a computer.

Overview

Timeline: 4 months
Role: sole researcher and designer
Goals: Improve efficiency and intuitiveness of planning class schedules to meet graduation requirements
Constraints: WebSTAC and WashU class registration

A bar graph showing how many pain points from my interviews and surveys fell into each category.

What problems do people face at WashU?

I sent out a Google Form and walked around campus asking people, “What problems or issues have you had at WashU?”

I found in-person interviewing to be far more effective, because while some people seemed uninterested, nobody declined to respond. Some people were very happy to talk to me about issues they’ve had.

I asked my dad to post the Google Form on the parents Facebook page, a popular place to discuss WashU problems, for parents to send to their kids, and that got a lot of responses.

Affinity mapping of the WashU communities pain points
Affinity mapping of the WashU communities pain points

Affinity Mapping

I sorted the pain points I collected into categories:

I considered which problem would be both interesting to address and possible to solve with a digital tool. While dining and transparency had the most responses, dining had just been addressed by the school, and transparency would take more than a digital tool to solve. Advising seemed like an interesting and solveable problem, so I decided to make it the focus of my next interviews. My classmate chose to focus on advising as well, so we pooled together our research.

 A list of possible users and a list of possible problems they might have with scheduling.

Users and Problems

Using the information I learned from interviews, I wrote down a list of possible users and a list of problems to solve.

Personas for two WashU students including a Freshman who wants to try a lot of different things and a Junior pre-med who wants to complete requirements and have time to study and volunteer.
Personas for two WashU advisors including a Sam Fox professor who enjoys advising and an Economics professor who doesn't enjoy it.

Personas

I created two student personas and two advisor personas for potential users of the website.

One student is an undecided freshman in Arts & Sciences who wants to try a variety of classes and clubs. The other is a pre-med Junior who wants to complete all the required classes on time with good grades, with time to study for the MCAT and volunteer.

One professor teaches Graphic Design and enjoys advising but struggles to help students in majors she isn’t that knowledgeable about. The Economics professor cares deeply about teaching economics but not so much about advising, as he finds it time consuming.

I used these personas to think about what features would be helpful in a scheduling and advising tool. I also came to the conclusion in class that a digital tool could eliminate the need for schedule advising.

Quotes from people I interviewed about which project ideas they preferred.
Bar graph of peoples responses to which idea they liked best.
A mockup of AI advisor schedule editing on a computer.

Feedback on Ideas

I wrote down four ideas for digital tools that could address the problems we came up with, and printed them out on a piece of paper. Then I walked around and asked people which idea they thought would be the most useful.

This method of interviewing got the most positive responses overall. People seemed the least bothered by it and could just pick one without having to say much. A lot of people did have additional helpful comments though.

“Schedule Sandbox”, a tool where you can drag and drop classes into a four year schedule, was the winner.

Screen shots of two AI scheduling  websites I researched including Motion and Teamup
Screen shots of three AI Chatbot websites I researched including Intercom, Ada, and Langchain

Competitive Analysis

I researched several tools for building an AI chat bot. I learned that they could respond based on support content and custom answers, and understand complex content without training. This seemed like the perfect solution to problems with advising. I also found two AI scheduling tools that prioritize tasks and work-life balance, optimize schedules, and work for individual use or with a team.

An early user flow map.
User flows and journey mapping.

User Journeys

I mapped out the user journeys of the earlier student personas by task and with potential wireframes, considering what features and layout would be ideal for the users’ needs.

Crazy 8 exercise.

Crazy 8 Exercise

I did the crazy 8 exercise a few times to create a variety of layouts for mobile and desktop for the features I came up with in my user journey mapping. This was helpful for thinking about how the different areas of the page would interact and how the user would navigate the site.

Low fidelity wireframes.
Low fidelity wireframes.
Low fidelity wireframes.

Wireframes

I chose some of my sketches to develop further into low fidelity wireframes. I used my personas, journey maps, and user feedback to think about which layouts would best meet users' needs for planning a long-term schedule with confusing requirements.

Mood board for AI Advisor.
Competitive analysis research on color palettes and visual language.

Visual Design

I made several mood boards including one that felt academic and friendly, one with the WashU site color palette and typography, and one based on my competitive analysis.

Cover slide of my initial research presentation

User Testing

We went as a class to visit the UX Research Methods class, presented our research, and recieved comments from members of the class after they went through our prototypes. The class gave feedback on what confused them and what felt intuitive as they went through the prototype. They also gave suggestions for visual changes that would make the function of different features clearer, such as color variation, more microinteractions, and icons.

Figma prototype frames.
Figma prototype frames.

Prototypes

I made iterations of mid-fidelity prototypes based on the user journeys, personas, and user testing. Throughout this process, I considered the relationship between the AI advisor, schedule, and progress tracker and how this would impact the layout. I also broke down the user journey into onboarding, schedule building, and schedule editing.

Final Prototype

After the initial testing, I expanded the prototype to include more tasks and alternated between making changes to the layout, interactions, and features and asking for feedback.

Figma Prototype