Grace
AI assistant connecting students with campus ministries
Overview
-
Our team was tasked with creating an AI assistant for the Faith Forward sector of Called. Faith Forward helps high school seniors connect with college ministries.
-
Emmy Snipes, Product Designer
Brandon Livingood, Director of Product
Jason Brown, Engineering Manager
Ben Musil, Frontend Engineer
Mike Bostone, Mobile Engineer
Nick Paul, Java Developer
Brian Manson, Backend Engineer
Peter DeVita, Backend Developer
-
Product Design
-
We had a short runway to get the interface and user flow ready for an international conference demo, which kept things focused. The saving grace was that a lot of the LLM and community-mapping work had already been done. Earlier AI development had been paused to make room for other product priorities, so that groundwork was just waiting to be picked up. This also unfortunately meant that we only had time to test the prototype internally. There simply wasn’t time to gather test users and conduct any interviews or usability testing before the conference.
User journey mapping
Students access Grace through a web link and are walked through a series of questions about their college plans, concerns, and interests. Then, Grace matches them with a campus ministry that fits. The primary challenge were figuring out when to present prompt suggestions so the experience felt guided rather than pushed, since finding a faith community is a personal decision for a lot of students.
Conversation design
Conversational AI can be a pain. Unlike a standard form, there's no obvious next step, just an open field and the pressure of not knowing if you're asking the right thing. We wanted to take that pressure away by being upfront about what Grace could actually do, and making sure that first interaction felt less like a test and more like a conversation worth having.
Visual design
I owned the visual design end-to-end. The starting point was Called's existing design system, but Grace needed to feel like its own thing. I used the existing color tokens to build out atmospheric gradients, then brought in glassmorphism and subtle motion to give it some depth, while maintaining consistency with the existing app.
