AI-native, mobile-first chatbot

This new post-RFI product was built to offer personalized answers to common enrollment questions prospective students have but may not feel comfortable speaking with an admissions representative about. Additionally, all conversations can be accessed by Admissions to help them provide a more tailored approach to supporting each student. Launched May 2026.

core responsiblities

As the sole UX Designer, I led end-to-end UX strategy and design of this new product.

My process

early usability testing & conversation design

I created med-fidelity mobile prototypes and conducted a series of usability studies based on early product requirements that helped guide the product strategy, ensuring user needs and expectations were accounted for before moving into the UI design phase.

The first series of tests focused on what would help engage users in conversation, observing their behavior, and getting a better idea of their overall expectations and preferences.

Do users prefer to start the chat with a few “get to know you questions” before getting into content topics, or would they prefer to go right into exploring FAQs?

Do users prefer having the primary CTAs fixed at the bottom of their screen, or do they prefer having a few external resource links to explore at the bottom with CTAs moved into the menu?

key findings

70% of testers preferred choosing a question from the FAQs vs. being asked a few “get to know you” questions at the beginning of the chat.
It’s beneficial to have options available for users to click on as well as having a more freeform chat available for those who want a more personalized chat experience.
Having the “Apply Now” button visible (either as a button or clickable link at the bottom) was beneficial to most. They would likely click that, but if they had questions they would utilize the chat feature.

Next, we wanted to understand how users perceived the AI avatar as well as how valuable and trustworthy they find AI chatbots to be in general.

How do they feel about the AI avatar? Did they find the AI chat useful or would they rather talk to a real person?

Key findings

60% thought the avatar made the chat seem more approachable; 20% thought the avatar made them think they’re talking to a real person; 10% felt unsure about it; 10% said the avatar image meant nothing to them.
The Welcome screen was unnecessarily confusing and needed to go.
Overall, participants were skeptical of AI chatbots and felt they could be generic, limited, and make mistakes, which has led to a general lack of trust. However, some find AI chats to be helpful for general questions with the opportunity for a human conversation for more specific questions or problems. The AI chat needs to be non-generic, smart, personable, and “surprisingly helpful” in order to build trust.

design evolution

As we moved into the design phase, I researched human/AI interaction best practices and best-in-class chatbot examples. Then, brought these insights into a working session with the creative team, where I facilitated a collaborative brainstorm to define the chatbot’s interface and core components based on user, tech, and business requirements.

From there, I translated ideas into a series of iterative desktop and mobile mockups that I regularly presented to various internal teams to gather feedback and align everyone on a clear direction forward.

Tech requirements were continuously being defined alongside the design phase, so lots of collaboration was required to get it right.

edge cases

Continued to research best practices and provide mock-ups for transition screens and edge case scenarios as needed, such as the redirect message and various error states.
Provided mock-ups that rebranded the final template for the next school that will be white labeling this product.

I won a company-wide award for my work on this project (August 2025).