Waymo Airport
Your flight changed. Your ride should know.
A booked flight is already a plan for a ride.
One trip from flight to curb. Pickup changes stay in your hands.
Arriving at 2:32 PM
8 min
Phoenix Sky Harbor · Terminal 4
AA 1932 to San Francisco
Level 1 · South Door 8
Prototype scope and test measures
- Plan · ride · arrival
- 3 moments
- Suggested rides
- Accept rate
- Timing & trust
- Overrides
Interactive concept. Acceptance and timing corrections would be measured in a live pilot.
The process behind Waymo Airport
- Suggest, then confirm.
- A delayed flight can move the pickup. The rider chooses whether to leave later or keep the extra time.
- Keep the ride in view.
- Route and arrival time stay primary. Gate and boarding details open only when needed.
- Test trust first.
- Measure suggested pickups accepted and timing edits. Stale flight data should fall back to the rider’s saved plan.
Why this case study: Waymo is the consumer product I find interesting right now. Airport pickup is the part it hasn’t figured out, so I took a shot at it. This remains a self-directed concept; the prototype demonstrates the interaction model, not measured impact.
Product shape
One connected trip, three jobs.
Predictive concierge
Uses the itinerary, live flight status, and security waits to suggest timing without taking control away.
Connected journey
Carries context from calendar permission through the airport ride and into the arrival leg.
Demand sensing
Turns an unavailable destination into an explicit waitlist signal instead of a product dead end.
Decision chapters
Skim the calls. Open the ones you want to interrogate.
Decision 01
Predict the trip without taking the wheel from the rider.
Signal
The itinerary already contains the flight, while delays and security waits make a fixed pickup time stale.
Bet
A narrow calendar permission can remove scheduling work if every automated suggestion remains reversible.
Inspect the decision
Alternatives
Manual flight entry, a generic reminder, or fully automatic rescheduling. Manual entry repeats work; a generic reminder lacks context; full automation creates the largest trust failure.
Artifact
The Plan stage demonstrates a delayed flight, a suggested timing change, and an explicit pickup confirmation. Calendar permission is part of the broader concept.
Tradeoff
The concept adds a permission step and accepts more user confirmation in exchange for legibility and trust.
What changed
The prototype moves from an automatic-booking idea to proactive suggestions with visible overrides. This is a design conclusion, not a measured result.
Decision 02
Make the ride the calmest place to understand the flight.
Signal
After pickup, route, security, gate, boarding, and weather remain fragmented across apps even though the rider is already in a large-screen environment.
Bet
A glanceable in-car flight deck can reduce app switching while keeping driving information primary.
Inspect the decision
Alternatives
Keep the car screen navigation-only, mirror the airline app, or send everything to the phone. The prototype instead limits itself to trip-critical states.
Artifact
The Ride stage keeps the route and arrival time visible, with boarding context and expandable flight details. Security, weather, and onward travel belong to the broader exploration.
Tradeoff
More context can become cockpit clutter. The design uses tabs and progressive detail rather than showing every module at once.
What changed
The prototype establishes a hierarchy: navigation first, time-sensitive flight context second, optional exploration behind tabs.
Decision 03
Treat an unavailable city as a signal, not a dead end.
Signal
A rider selecting an inactive destination has unusually high-intent context: a real itinerary, city, and travel date.
Bet
A transparent waitlist at that moment creates a useful expansion signal without pretending service exists.
Inspect the decision
Alternatives
Hide unsupported cities, show a static availability message, or collect broad survey interest elsewhere.
Artifact
The original exploration included an unavailable-city waitlist. This remains a future extension; the current prototype focuses on the airport trip.
Tradeoff
Demand intent is not operational readiness; the signal may help rank research, but cannot decide a launch on its own.
What changed
The exploration treats no-service intent as a research input for expansion. No market-expansion outcome is claimed.
The broader exploration: onward rides and expansion demand. These diagrams describe the concept beyond the airport-trip prototype above.
The growth flywheel
Every airport trip feeds the next launch. The in-car moment converts riders where Waymo runs — and quietly measures demand where it doesn’t.
Every airport trip
A ride with a known flight, origin, and destination.
Two demand signals
Lock-in where Waymo runs; waitlist intent where it doesn't.
Smarter expansion
Verified demand ranks the next cities to launch — which creates more trips.
Where Waymo is available → convert
In-car, the onward ride is offered before baggage claim — the next trip is locked in.
Where Waymo isn’t yet → capture
The destination toggle logs verified demand, turning a “no ride” into the input for where to expand next.
The strongest objection
“Why build calendar + flight detection at all? Just notify riders when they land somewhere Waymo already operates.”
Because the calendar turns the ride tothe airport into an intentional check-in. It already knows the whole itinerary — flight, destination, even the hotel — so the inbound ride can pre-stage the outbound one: while you’re having a smooth trip to the gate, your destination ride is queued up, timed to how long Waymo expects checkout and customs to take, pre-filled and one tap to confirm when you land.
A landing-only ping can’t do that. It fires after you’ve landed, with none of the context to book ahead — and it only ever sees cities Waymo already serves, so it can’t feed expansion.
The bet: a great ride to the airport is the best moment to win the ride on the other end. Capturing both legs — plus the demand signal for cities that aren’t live yet — is worth the calendar permission in a way a cold landing-ping never could be.