Zone-based gate seating for Phoenix Sky Harbor International Airport, built around traveler choice.
Solo Designer
Fall 2025 to Winter 2026
Figma
Claude Opus 4.6
Google Gemini
Nano Banana
Smart IoT Environment
Smartphone
Three seating zones, Quiet, Standard, and Social, let travelers choose the kind of space they want
Overhead depth sensors and seat pressure sensors read the gate at zone level, never per person
Signage, lighting, and an optional app show where seats are open, stepping in more as the gate gets busier
Recover the roughly 40% of seats that sit empty but socially claimed
Remove the 6-step claiming sequence travelers perform today
The gate reads at a glance in person, with the app strictly optional
At a Sky Harbor gate, roughly 40% of seats end up empty but socially claimed while travelers stand along the walls instead. The cause is a ritual every traveler performs, a 6-step boundary-protection sequence: enter and scan the room, pick the safest-looking seat, claim the seats beside them with bags, put on headphones to signal they want to be left alone, then hold that boundary until boarding. The environment creates the problem, not the passengers.
SmartGate splits the gate into three zones, Quiet, Standard, and Social, so choosing a seat becomes choosing a space. Sensors read how full each zone is, and signage, lighting, and app alerts respond as the gate fills. The zones read clearly in person; the app adds detail only for travelers who want it.
A quick walkthrough of the gate in action: the three zones, the sensing that reads them, and the companion app that follows a traveler from choosing a space to boarding.
Solo, so I owned all of it:
The fieldwork: three observation sessions and three interviews inside Terminal 4
The contextual design work models, from the physical model to the nine identities
Two empathy maps, the passenger and the gate agent
The seven-stage journey map and the opportunity map
The three-zone environment and its signage, seating and lighting
The companion app, delivered as a clickable prototype
Just me, with Claude as an AI design partner
Fall 2025 research; ~6 weeks of design, Winter 2026
Three observation sessions and three semi-structured interviews inside Terminal 4, about three and a half hours of field time, open-coded into 40+ topics and built into contextual design work models. The gate moves roughly 50 million passengers a year, and not one of the three findings put the problem in the traveler.
About 40% of occupied seats held bags rather than people, and seats between strangers went constantly unused. Travelers judge whether a seat is available by the people around it, not by whether it is empty.
9 distinct traveler identities with opposing needs, from the Anti-Social to the People Person, confirmed that one uniform gate layout cannot serve every passenger.
The fuller a gate gets, the harder travelers work to avoid each other. The moment someone walks into a crowded gate and scans for a seat is where a connected system can help most.
“Me and the rest of the world are uncomfortable sitting here together.”
Research left one challenge to build against: take the boundary decisions off the traveler and give them to the space. Answering it took three moves.
I mapped two people, not one. The passenger map surfaced the anxiety underneath the claiming ritual: it spikes the moment a crowded gate comes into view, runs on thoughts like “is that seat too close? Will they think I’m weird?”, and leans on devices for privacy, with about 80% of travelers using one as a social shield.
The gate agent’s map found the other half of the problem. Their seat metrics say available while the room looks full, crowding is estimated by eye, and they absorb frustration over conditions they cannot control. Two people, one room, and no shared picture of it, which is why the traveler app and the gate operations feed were designed as one product.
The journey map followed a passenger through seven stages, from pre-gate to boarding, tracking emotion alongside behavior: anticipation on the walk over, an anxiety spike on entering, guardedness through claiming and signaling, fatigue by the time boarding is called. Friction peaks in the middle, surveying the room and settling in to wait, exactly where the environment offers no help. That narrow window is where SmartGate spends everything it has.
Seven phases from pre-gate to boarding. Emotion sinks into stress at survey and fatigue through the wait, and the highest-friction stretch, where the boundary work concentrates, runs from surveying the room to holding a seat until boarding.Everything the define work found compresses into a single statement, and every decision after it was checked against this one sentence rather than against taste.
What follows in Design is that sentence applied three times: to the room, to the signals the room gives off, and to the app that reads them.
The design focus applied three times: to the room, to the signals the room gives off, and to the app that reads them. Physical first and digital optional, because a traveler with a dead battery still has to find a seat. Zone renders generated with Nano Banana.
Three zones, each tuned to a cluster of the nine identities. Seating geometry, lighting and sightlines carry the meaning, so a traveler picks a zone by reading the room rather than by reading a sign. The archetypes under each zone are the ones whose asks that geometry answers.
The zones are legible when the gate is calm. The harder case is a gate filling up, which is exactly when a traveler is scanning and least able to judge. Sensors read occupancy, and the signage and lighting update so the room reports its own state.
The smart environmentThe environment itself went through three passes: I mapped the full system onto the gate first, stripped it back to a clean render, then revised the space after design feedback. Base renders generated with Nano Banana.
A step back from the renders, this is the logic that runs the gate: the room works for everyone by default, and the app is optional detail. Sensors read each zone's social occupancy, and the environment shifts with the crowd, staying passive while a zone is quiet, turning active as it fills, and converting the zones into boarding groups once boarding is called. The same feed powers three tiers: the physical room for everyone, the app for travelers who opt in, and an operator dashboard passengers never see.
Everything above works with a dead phone. The companion app is for travelers who want more than the room can say at a glance: which zone has space now, where their people are, and when to move for boarding. The gate agent sees the same picture from the podium.
SmartGate lives inside Sky Harbor's own brand, part of the airport rather than another app to download and learn.
Gate B22 at a glance: every zone shows a photo of the actual space, a live seat count, and a plain word like Spacious, so travelers can choose before they walk in.
One zone up close: the ambient features, live availability, and a seat map that deliberately shows estimates, never assigned seats.
Preferences persist: save Quiet once, and on every future trip through Sky Harbor the app flags when it has space.
The overview once a zone fills: Quiet reads Full and offers to watch it, Social shows ten seats free. Nobody has to walk the gate to learn that.
Zero seats is information, not a dead end: the app says Full honestly and offers to watch for an opening.
The clearest proof the app is optional: boarding, gate, group, seat and how each zone is emptying, all without unlocking the phone.
Boarding closes the loop: group progress, a live passenger count, and the zones clearing out as the gate empties.
The app reports exact figures, 0 of 28, 10 of 24. The research finding underneath this whole project is that physical and social availability are different numbers, and an exact count of empty seats is the gate agent’s problem restated: data says available while the room looks full. The count needs to exclude bag-claimed and between-strangers seats, or it will be distrusted for the same reason.
The operator empathy map names passengers with mobility needs struggling without help, and none of the three zones answers it. A fourth zone, or a reserved band inside Standard with clear circulation and companion seating, is the obvious gap in a design that otherwise claims to serve every identity.
Quiet, Standard and Social are distinguished largely by hue in both the signage and the app. Add a second cue that survives color blindness and low light, in shape, icon, or seating silhouette, so the zone reads without the color.
The screens use words in one place (Spacious, Filling, Full) and precise counts in another. Both are defensible; using both at once teaches a traveler two mental models for the same fact. Choose the words, and keep the number as the detail underneath.
A live, clickable Figma prototype of the SmartGate companion app: choose a zone, save it as a preference, and follow the flow through to boarding.
Every pain in the research traced back to the space itself: fixed forward-facing rows, clustered outlets, zero wayfinding for social availability. Reframing the gate as the thing being designed, rather than the passengers being managed, unlocked the whole concept.
Gate agents watch seat metrics say available while the room looks full, and they absorb the frustration for conditions they cannot control. The gate operations feed exists because an environment that helps passengers while leaving agents guessing solves half the problem.
The zones are designed to speak for themselves; the app adds detail only for travelers who want it. In a space where many people never open an app, the room itself has to be the primary interface.
Claude supported design feedback, system flowcharting, and rationale refinement; Nano Banana generated the base environment renders. The research, the problem framing, and every design decision stayed mine.
The clickable prototype covers the companion app, but the core bet is spatial. I’d test zone legibility with a scaled physical mockup before touching another screen.
Each zone serves named archetypes from the research, and that mapping is a hypothesis. I’d recruit travelers across the 9 identities and watch which zone they actually choose.
Zone-level sensing avoids tracking individuals by design, but the environment never tells travelers that. A system reading a room full of people should say what it senses and what it never stores.