ROLES
Design Consultant
UX & Interaction Design
Conversational AI Experience Design
Visual Design
Branding
SCOPE
Collaborate with an MIT CSAIL researcher to translate conversational machine-learning technology into an intuitive mobile experience for tracking nutrition through natural language.
OUTCOME
Designed the original consumer experience for a conversational AI nutrition app that launched publicly and remains available today. Later reimagined the experience around modern AI interaction patterns, with the resulting design now used in undergraduate AI engineering lectures at Loyola Marymount University.
Making artificial intelligence feel human
In 2018, conversational AI was still largely confined to research labs and emerging consumer applications. At MIT's CSAIL, researchers had developed technology capable of interpreting natural-language descriptions of food and translating them into nutritional information.
The challenge was making that capability useful to everyday people.
Traditional calorie-tracking apps required users to search databases, select foods, estimate portions, and manually log their meals. Coco offered a different possibility: what if tracking nutrition were as simple as telling someone what you ate?
Our task was to turn that technology into an experience people could understand, trust, and enjoy using.
2018 / Designing for conversational AI
I developed a messaging-based interface that allowed users to describe meals in their own words, replacing much of the structured data entry typical of nutrition apps.
The underlying AI could identify foods and quantities, retrieve nutritional information, and calculate calories and macronutrients. The interface translated those interpretations into a clear conversational exchange.
One important decision was making the AI's understanding visible. Rather than presenting only a calculated result, Coco displayed the foods and quantities it had identified, using color-coded text to distinguish key information.
This made the system's reasoning easier to follow and gave users an opportunity to recognize incorrect interpretations.
The goal wasn't simply to make food logging faster. It was to make an unfamiliar technology feel predictable and trustworthy.
2024 / Revisiting the original design
Years later, I returned to Coco to explore how the interface could be updated without changing its underlying interaction model.
The first exploration preserved the original navigation and conversational workflow while modernizing the visual language.
I refined the typography, spacing, color hierarchy, and presentation of nutritional information to create a more polished, approachable experience.
This exercise highlighted an important distinction: a product's visual design can evolve considerably while its underlying interaction architecture remains unchanged.
But it also raised a more interesting question: knowing what we know about conversational AI today, would I design the experience the same way?
2026 / Rethinking the experience for today's AI
My second exploration went beyond visual design to reconsider Coco's interaction architecture.
The original application combined a conversational interface with conventional navigation, reflecting familiar mobile design patterns of the time.
We no need longer need to dip users' toes into AI. Today, they're more familiar with interacting with AI, so I'd make conversation the primary interface.
I removed the persistent tab navigation and organized the experience around a single conversational canvas. Nutritional summaries, historical data, and account features moved into a supporting interface that could be revealed when needed.
The result was a simpler experience that placed the AI interaction at the center, using progressive disclosure to make additional functionality accessible without competing for attention.
Rather than designing a traditional nutrition app that incorporates conversation, I explored what it means to design an AI-first product from the ground up.
Outcome & Takeaway
Coco gave me an early opportunity to bring emerging AI research to consumers, designing the interface for a publicly launched product that made natural-language nutrition tracking accessible to everyday users.
Years later, my reimagined design is being used in undergraduate AI engineering instruction at Loyola Marymount University, demonstrating how thoughtful interaction design can make complex AI capabilities approachable.
The project reinforced something that remains central to my approach to emerging technology: as AI becomes increasingly powerful, the challenge isn't simply giving people access to its capabilities.
It's designing experiences that make those capabilities understandable, trustworthy, and genuinely useful.