Dietary Filter • Concept 2026

A Google Maps filter that finds food safe for your diet

Din Tai Fung, Urban Plates, and Siam Corner demonstrate Verified, Limited info, and Unknown with provenance and actions.
Role
Product Designer, UX Researcher, Design Systems
Timeline
April 2026 to July 2026 (class quarter, then a solo three-week v2)
Team
Mason Oelschlager, Asher Hardy, Rayleen Marquez
Tools
Figma, Figma Variables & Components, Material Design 3, FigJam, Claude Design
Live
Concept
Second-round testers with real allergies, halal and keto diets
5
High-fidelity directions tested into one hybrid
2 → 1
Of Google listings that fill in any allergen field at all
13%

A dietary filter for Google Maps, for people with food allergies. My class team designed it, then I kept going alone for three weeks. It's a Figma concept, not a shipped app.

My interviews found the hard part is social. People hide their allergy so they won't be a burden.

Problem

For someone with an allergy, picking a restaurant is homework done outside Maps.

One woman I interviewed checks an allergy site, then Reddit, then Google reviews before any new place. None of it happens in Maps.

She has a serious peanut allergy and eats out in groups. She's the only one tracking what's safe.

Maps today, then the same search with the filter.
Three interview quotes describe group dining and the participant receiving someone else’s restaurant choice.
In her words: the group picks, and she copes quietly.

Process

"Likely safe" was the most dangerous thing I designed.

The class version had a state called "Likely safe." It tested well. But for someone always on guard, a soft green light says relax. That's how mistakes happen.

So I made three states, each with its own action. Verified means go. Limited info means be careful. Unknown means you're on your own. The rule: clear slowly, warn quickly.

Each state says where its answer came from.

Process

My model called peanut Pad Thai Verified, so dishes got a Contains state.

Pad Thai that lists peanuts is fully known and fully dangerous. My model only measured how much we knew. So it showed a green Verified chip. Exactly backwards.

So dishes got a fourth state, in red: "Contains peanuts."

Before, Verified. Now, Contains peanuts.
The Menu tab separates Flagged for you, including Contains peanuts, from No flags found dishes.
Dishes get a fourth state, named for what's in them.

Process

I cut the signature feature because my participant doesn't pick the restaurant.

The class version let you share a place without saying why. But a normal Maps share already says nothing about allergies. And she doesn't pick the place: "Yeah, it was Lucy."

So I flipped it. When a friend shares a place, the app checks it against your profile. If it doesn't work, it names two nearby places that do.

The cut outbound share beside an inbound restaurant recommendation with two nearby alternatives.
Left, the cut share. Right, this place is wrong, these two work.

Final Design

Three states for a restaurant, four for a dish, one profile behind both.

You set your profile once, even partial limits. Then every place and dish in Maps reads against it. None of it tells the group why.

Reflection

Still a Figma concept. Next, I'd ask someone with an allergy if it's true.

The hardest part isn't the screens. US restaurants don't have to list allergens, and most Google listings don't.

I haven't tested the trust model with anyone yet. That's the next thing.