Pairfecto
CLIENT
MIIPS students at CMU Silicon Valley
TIME
4 weeks (ongoing)
SKILL
UX Design, Visual Design, Prototyping
TOOL
Figma, Notion
MY ROLE
Product Designer
TEAM
Product Manager + Developer
An AI-powered app that learns and remembers both you and your partner’s restaurant preferences, making it effortless to find the perfect dining spot for any occasion.
I took on the role of the sole designer, responsible for shaping the UX flow, UI, and brand identity, enabling the team to move into development and meet MVP submission deadlines for startup competitions.
(The project is currently in development.)
PROBLEM
Defining User Needs
Choosing a restaurant for a date can often be a frustrating and time-consuming task, especially for those responsible for making the decision in a relationship.
Through desktop research and 5+ user interviews focused on young adults on the West Coast, particularly in Silicon Valley, the team identified this as a common pain point—one that frequently falls on boyfriends.
Narrowing Down to One Product Idea
Recognizing this need, the team saw an opportunity to develop an app that matches users’ dietary preferences and restrictions while delivering personalized restaurant recommendations. Leveraging a Large Language Model (LLM) and a continuously learning system, the app adapts to individual tastes by memorizing past selections and refining suggestions over time.
In short,
the team is building an AI-powered app that understands and remembers your partner’s restaurant preferences.
Below are the features the team aims to include in the app.
Framing the Problem Statement
By synthesizing all available information, including the spec, prior user research, and competitor analysis from the team, I reframed the problem statement to highlight the key value propositions:
How can we emphasize personalization and simplicity, while tailoring the content to couples to differentiate from other restaurant-finding apps?
PROCESS
In response to the problem statement, I identified key design challenges and made informed design decisions based on insights from existing research documents and my own research conducted throughout the process.
UI Decision
Mood board Collection
Distilling key attributes from the target audience—male software engineer boyfriends on the West Coast—and potential brand values, I began by exploring visuals that align with these listed themes:
#simple
#code/data-driven
#friendly
#futuristic/AI-driven
Narrowing Down
I was inspired by a conversation with a friend, who also happens to be a target user, which helped shape the concept I ultimately landed on.
His initial reaction to “I’m working on an app that helps boyfriends find restaurants for their partners” was:
"Maybe it needs some psychic abilities."
As I laughed out loud, the word “psychic” stood out to me, ultimately shaping the core concept of my UI decisions. Here are my 10 explorations surrounding the concept.
Here are the final decisions, informed by feedback from the product manager and several target users.
Design System
I aimed to blend the AI-powered, futuristic theme with dark gradient colors on the splash screen and onboarding pages, evoking a sense of mystery. As the home page unfolds, the UI transitions to a brighter palette, symbolizing the moment of discovery—emerging from uncertainty into clarity.
Design
Challenge #1
How might we enhance personalization through effective profile creation flow?
Precedent Study
First, I conducted desktop research on precedents that incorporate profile creation during onboarding. I explored dating apps like Bumble BFF, which not only include a questionnaire but also effectively communicate the purpose behind collecting user data.
Wireframe
Building on precedent studies, I sketched the key frames of the onboarding process, ensuring a smooth and intuitive user experience.
Key elements include:
Progress bar to indicate completion status
Selection tags for personalized input
Return & Next buttons for easy navigation
Iterations
The first change I made was adjusting the UI of the selection box to create a clearer distinction between the default and selected states.
Another key change was adding a skip button—an essential option for users who may not want to configure their settings immediately.
Design
Challenge #2
How do we differentiate ourselves from apps like Google Map and Yelp?
Wireframe
I sketched the key frames of the search function, mapping out the user journey from the search homepage to detailed result pages.
Key elements include:
Search bar for easy input
Search hints based on history & trends
List of results for quick browsing
Filter & sorting options for personalized refinement
Image-based results for visual appeal
AI-generated tags derived from customer reviews
Integrated Google Maps reviews for credibility
Share button to easily send restaurant options to partners
Iterations
By gathering feedback from other designers and discussing feasibility with the product manager, I made the first major change to the arrangement of the result cards.
I presented three layout options: small cards in a grid, a list format, and larger cards displaying more information—similar to Google Maps.
We ultimately chose the larger card format, as reducing the number of clicks needed to access key information aligns with our user-centered approach.
Another iteration focused on refining the information displayed on the cards. By adding tags for special occasions, the recommendations better align with users’ intentions while also creating opportunities to partner with local businesses for featured placements in the News Paper.
Here's how it looks on the result page.
SOLUTION
Pairfecto is an AI-powered app designed to help you find the perfect restaurant based on your partner's and your preferences and dining habits.
# Profile creation
To get started, create a detailed profile by adding your key dates, dietary preferences, restrictions, and favorite cuisine type.
# Integrating Your Partner's Dining Preferences
Create a detailed profile for your partner. Pairfecto will remember both of your dietary restrictions, favorite cuisines, and special occasions, like anniversaries.
# AI-powered Result
By integrating user profiles, Pairfecto leverages real-time data from Google Maps and AI-driven insights to recommend restaurants by cuisine, ambiance, price range, and couple-friendly features.
By continuously learning from your choices and post-meal feedback, Pairfecto refines recommendations over time, eliminating indecision and making date planning effortless.
REFLECTION
Feasibility v.s. User Experience
Working with a software engineer on a fast-paced development timeline—especially with an impending online campaign—made feasibility a recurring topic of discussion.
As a designer, one key takeaway was the importance of advocating for users. Rather than compromising a feature simply due to time constraints, we can set milestones for iteration and future updates. Designers should serve as the voice of the users, ensuring that their needs remain the priority throughout the development process.
If there is more time…
If given more time, I would refine the layout configuration to ensure a smoother transition to the software developer.
For this project, we decided to utilize Figma’s Dev Mode, allowing the developer to directly access HTML and CSS code. With additional time, I would focus on fine-tuning margins, gaps, and other design details to streamline the handoff, reducing the need for developers to manually standardize the code.
minjeanchu@gmail.com