American Airlines

A connected journey.
A clearer next step.

Exploring contextual AI guidance for the moments between planning and arrival.

I reimagined the travel journey through AI-powered experiences, conversational AI, and intelligent interfaces—connecting trip planning, airport navigation, and arrival through a contextual AI Companion. The interactive prototype brings this vision to life, with traveler outcomes yet to be validated.

View the experience Try prototype
Contribution
Experience strategy, interaction design, visual design & prototyping
Scope
AI-powered planning → airport → arrival
Status
Independent concept · scripted demo
THE STORY IN 30 SECONDS
01
The problem

Travel questions change. Trip context should stay visible.

02
My decisions

Three journey states with relevant starting prompts.

03
Working experience

Nine prompts and scripted guidance around one trip.

04
Validate next

Can travelers act correctly and recognize the limits?

Where AI could help the traveler

Sample trip data · scripted responses
01

Before travel

Help me prepare

“What should I do before I leave?”
Trip detailsPreparationChecklist

Bring trip details and preparation prompts together so the traveler has a useful place to start.

02

At the airport

Guide my next step

“What should I check right now?”
Travel momentGuidanceConfirm

Surface relevant gate, timing, and airport questions, with a reminder to confirm current details.

03

On arrival

Help me keep going

“What comes after landing?”
ArrivalSaved plansReturn trip

Continue the conversation into destination plans and preparation for the return journey.

04

Across the journey

Keep the trip connected

“Can I get help without starting over?”
Same tripNew questionRelevant help

Keep the sample itinerary visible while prompts and scripted guidance change with the travel moment.

02 / THE EXPERIENCE

One trip. Three moments of guidance.

Three design choices to explore in the preview. Next, I would test whether travelers find the right action and know when information needs confirmation.

  1. 01

    Prepare for the trip

    Relevant prompts provide a starting point.

    The sample trip stays visible while the companion offers a short planning response.

  2. 02

    Find the next step

    Guidance shifts. The trip stays in view.

    Prompts shift toward gate, timing, and airport questions. The response asks the traveler to confirm information.

  3. 03

    Continue after arrival

    Support continues beyond the flight.

    The conversation turns to saved plans and the return journey, completing the illustrative sequence.

AI COMPANION / INTERACTIVE EXPERIENCE

Explore your connected journey.

American AI NavigatorOpen full screen

03 / COMPETITIVE ANALYSIS

Six perspectives. One connected vision.

Choose an airline to explore the insight behind my design.

THE LANDSCAPE Choose a signal
American / THEIR STRENGTHSource

Trip essentials, together

Check-in, boarding passes, changes, and disruption support in one trip hub.

MY APPROACH

Start with the traveler’s moment

Keep the trip visible. Adapt the questions to planning, the airport, and arrival.

Test next

Can travelers find the right next step?

Try the concept

EVIDENCE → DESIGN OPPORTUNITY

AI maturity: current evidence and future opportunity

Build on trip tools. Validate grounded guidance before adding actions.

Research snapshot · Oct 27, 2026
CAPABILITY EXPLORERExplore the evidence. Follow the design decision.

01 / Choose a capability

02 / Compare with

Disruption & connections

Both sit at level 2: contextual. Their approaches differ—see the evidence below.

FIXED BASELINE

American

Reported available
  1. 1Assistive
  2. 2Contextual
  3. 3Grounded conversational
  4. 4Confirmed transactional
  5. 5Bounded orchestration

ML helps surface rebooking choices for disrupted trips.

AI & rebooking ↗Booking & trip tools ↗
SELECTED COMPETITOR

Delta

Available · beta
  1. 1Assistive
  2. 2Contextual
  3. 3Grounded conversational
  4. 4Confirmed transactional
  5. 5Bounded orchestration

Concierge explains disruptions and helps find alternatives.

Delta Concierge ↗
03 / APPLY TO MY CONCEPT

My design opportunity

Keep recovery options and trip context together, then guide the next question through contextual prompts. Live recovery options remain a future exploration.

What I would validate

Can travelers choose an appropriate next action and know what needs confirmation?

Try the current prototype

Levels describe capabilities, not overall airline quality. Unknown means insufficient evidence.

Sources checked: Assessment reviewed: Company/partner claims · not independently tested

MY PROPOSED SCENARIO

24-month opportunity

Oct 27, 2026 → Oct 27, 2028

Planning assumptions to validate with engineering—not American’s roadmap or a forecast of competitor performance.

0–6 months Oct 2026 → Apr 2027
Validate the next step

I propose testing the three existing journey states and contextual prompts before adding live AI.

Requires: Scenario rubric, representative travelers, trip-data mapping.

Gate: Exit: record correct next actions, assistance and comprehension; establish baselines.

Uncertainty: Relevant prompts may still leave travelers unsure what to do.

6–12 months Apr → Oct 2027
Ground the guidance

Explore connecting the existing conversation pattern to authoritative trip information, with sources and handoff.

Requires: Identity, consent, read-only trip access, freshness checks and response evaluations.

Gate: Entry: validated scenarios. Exit: pass agreed accuracy and escalation criteria before a limited pilot.

Uncertainty: Stale or incomplete data can make helpful-sounding advice wrong.

12–24 months Oct 2027 → Oct 2028
Evaluate one confirmed action

Only after validation, explore a traveler-confirmed handoff from guidance to an eligible airline action. This is beyond the current demo.

Requires: Booking rules, inventory, permissions, confirmation, audit logs and recovery support.

Gate: Entry: validated grounding. Exit: demonstrate authorization, failure recovery and human escalation in a sandbox.

Uncertainty: Integration and eligibility may make this scope infeasible; defer autonomous multi-step recovery.

My priority: keep trip context visible and test the next useful action. Build on existing recovery tools; defer autonomous recovery until permissions, reliability and escalation are demonstrated. Measure: correct next-step completion first, then guidance comprehension and time to a correct action. Record incomplete attempts; establish baselines before targets.

04 / DESIGN DECISIONS

Helpful guidance. Clear boundaries.

Two design choices, their limits, and what to test next.

01

Help for each travel moment.

Design
Relevant prompts. One shared itinerary.
Limit
Preset phases and prompts; no open-ended chat.
Test next
Can travelers find the right next step?
Explore the phases
02

The traveler stays in control.

Design
Guidance with clear sample-data labels.
Limit
Scripted replies cannot verify live travel details.
Test next
Do travelers know what to confirm?
Try the companion

05 / OUTCOMES & NEXT STEPS

What I built. What I’d test.

Built · sample prototype

One trip. Nine ways to start.

Three contextual prompts at each travel moment.

Before travel3
At the airport3
On arrival3
Each block = 1 prompt 9 total
Explore the prototype

What to try

Prepare
Choose a planning prompt and review the reply.
Navigate
Switch to the airport. Notice how the questions change.
Continue
Open arrival. See how support extends beyond the flight.

The design choice: one trip stays visible while the guidance adapts to the moment.

Published research · external

Connected journey

Can travelers find the right next step?

Want combined mobile travel capabilitiesIATA’s 2025 passenger survey: smartphone wallet, passport and loyalty cards for booking, payment and airport processes. Preference data, not task success or proof of this concept.IATA · 2025 ↗

What I would test next

  • Choose a relevant prompt for the travel moment.
  • Switch phases and locate the same trip details.
  • Identify a useful next step without moderator help.

Count correct task completions out of all eligible attempts.

Published research · external

Guidance clarity

Do travelers know what to confirm?

Concerned about inaccurate AI informationU.S. adults highly concerned; 5,410 surveyed in August 2024. Attitude data, not a clarity score.Pew Research Center · 2025 ↗

What I would test next

  • Explain the suggested next step in their own words.
  • Recognize sample data and scripted responses.
  • Identify which details need official confirmation.

Count travelers who correctly explain the advice and its limits.

Study results pending. Real participant data will be added after testing: study date, participant count, task completion (n/N), guidance comprehension (n/N), observed issues, and design changes. Published research above is external context, not a result from this prototype.