Creative AR & AI Travel Buddy App
Solo travel breaks down in the space between plans, where a traveler needs one clear next step and gets a page of search results instead.
TRA-BU is an AR and AI travel buddy that answers in the moment, points the way on the screen ahead, and stays present in between.
It is built for someone tired, unsure and already walking, so every answer arrives as something to do rather than something to read.
ROLE
AR Interaction & Spatial UI Designer, Team of two
Timeline
January to May 2024
Contribution
Spatial UI Design (AR),
Immersive Interaction Design,
Visual Language Development,
3D Asset Planning & Prototyping,
Narrative-Driven Design
01 | overview
Key highlights
Trip planning is treated as a task to finish, so tools compete on more options for a traveler who already has too many.
TRA-BU treats it as company instead, a guide that reacts in real time and turns the traveler's next decision into a direction to follow.
YOU ARE NOT ALONE
Human Like Travel Support
Plans break, and on a solo trip nobody else is there to fix them.
TRA-BU replies in real time, in a voice that reads as company.
TURN ROUTES INTO MOMENTS
Playful Guidance
Nobody looks forward to the walk between places.
AR arrows, characters and small games put something there.
NO DISTANCE REQUIRED
Travel For Everyone
Time, money and physical condition decide who gets to travel at all.
AR carries part of the experience to the person instead.
02 | concept
background
An unfamiliar place turns into a stream of small decisions, and each one costs more when there is nobody to ask.
Travel tools answer those decisions with more information, which is the one thing a tired traveler cannot process.
So the gap is not knowledge. It is having something beside you at the moment the decision arrives.
Tools are built to inform. None of them are built to accompany, which is where the question started.
💭 What if travel guidance felt less like using a tool, and more like being accompanied by a dependable virtual buddy?
Competitive Analysis Highlights
Two kinds of products already sit near this space, and each holds only half of it.
Gamified apps hold attention but know nothing about where the traveler is standing, while travel chatbots answer well and leave nothing to feel.
Neither one accompanies, which leaves the moment of walking and deciding unclaimed.
Playful, but not built for travel support.
Useful for travel questions,
but not playful or immersive.
Insight
The opening is the overlap between them, engagement on one side and real-time utility on the other. TRA-BU takes that position, useful in the moment and worth spending time with.
User interview
The competitor gap said what to build. It did not say where a companion would actually fail.
I interviewed three solo travelers about when travel drains them and how they use AI today. All three wanted the company. Each named a different reason they would not trust it yet.
Overseas Traveler Yeonoh Eo | 20 | University Student (Biology Major)
"When traveling, I stay more focused and understand places faster with an instant, interactive guide than with a detailed planner."
Problem
Detailed planners lose most of their value the moment the trip actually starts.
Insight
Support has to arrive in the moment, not in advance.
Domestic Traveler Jeagyu Jang | 23 | University Student (CS Major)
"Actually, I am skeptical about AR chatbot apps because they are hard to understand and use right away."
Problem
AR chatbot apps ask users to learn the interaction before they get anything back from it.
Insight
A companion has to be understood in the first minute, or it is never used at all.
Student Studying Abroad Seoyeon Park | 18 | University Student (CS Major)
"I do not often feel the need to use chatbots in situations where safety matters, such as traveling, because for now, Googling feels more trustworthy to me."
Problem
In safety-sensitive moments, an answer that cannot be checked is worse than no answer.
Insight
Trust depends on how fast guidance can be confirmed, not on how confident it sounds.
research summary
Solo travelers do not want a better plan. They want guidance that arrives in the moment and asks little of them.
Across interviews and competitors the failure repeated in the same place, before the value is ever reached, in the first interaction and in the moment trust is needed. So the goal became a companion that explains itself immediately and stays easy to verify.
Pain point
Learning the interaction before it gives anything back
Answers with no fast way to check them
Help that exists before the trip but not during it
Opportunity
A first exchange with nothing at stake
A source behind every recommendation, one tap away
Asking out loud without stopping to type
03 | development
Shaping the Build
overall project direction
The concept had to survive contact with a real body in a real place. I sketched the core interactions and character, built early 3D assets in Blender, and tested on myself until scale and interaction distance felt right. Those two variables were settled before a single screen was designed, because in AR they decide everything after.
Persona & Scenario
User Flow & Wireframe
Usability Testing
ideation sketch
The character had to read as a companion and as something worth collecting. Floating bird and fairy forms gave it presence without a body that blocks the street in front of you.
Persona & Scenario Map
Assaf books his first solo trip after a friend tells him how good it is, and is anxious about doing it alone.
Assaf Miller
23 • Introverted Traveler & International Student
Assaf hears from a friend that solo travel is a lot of fun and decides to go on a trip by himself. However, he is anxious about traveling alone.
Scenario 1
Book the Ticket
Assaf asks TRABU for cheap flights and a New York hotel, gets one recommendation with a link, and books it.
Book the Ticket
Scenario 2
Find the Gate
At the airport he asks where his gate is, answers two questions about his flight, and follows an arrow on the screen to it.
Scenario 3
Catch a Character
Walking with nothing planned, he finds a new character, captures it, and makes it his guide.
User flow
The flow had to be buildable, not just plausible. I checked each interaction for feasibility with an expert and cut what the technology could not hold up on a real street. What survived is a loop that answers, points and follows without asking the traveler to learn it.
Wireframe
Wireframes settled the paths before anything was styled. The open question was how much work the character does and how much the screen does. Resolving that in gray is what made the visual pass an execution step instead of an argument.
Usability testing
Two moderated rounds ran the same core tasks, first on the low fidelity prototype and then on the revised build.
1st Usability Testing
Everything that broke, broke before the companion had done anything for them.
Test Target
3 participants (university students, 20–25), 10 - 15 min each
Test Result
Worked Well
The idea felt creative and fun
Pain Points
The initial character setup flow felt unnatural
Privacy concerns emerged around location-based services
Design Changes After Testing
Simplified the character setup flow
Added a clearer privacy consent step during sign-up
2nd Usability Testing
What was left was not a usability bug but the interface saying the wrong thing about what TRABU is.
Test Target
Same participants as 1st UT, 15 min each
Test Result
Worked Well
Most tasks were completed smoothly
Pain Points
The top-left profile photo button felt too similar to dating app UI
The map icon was not intuitive enough at first glance
Design Changes After Testing
Removed the profile photo button
Redesigned the map icon to better communicate navigation
Usability testing
Then I conducted tasks. Tasks ran from easy to hard, opening with launch and sign up. The easy set exists to confirm the basics hold, since a failure there invalidates everything measured after it.
Easy
Moderate
Hard
First, the easy tasks focused on evaluating the basic sign-up process after launching the app. The goal was to confirm whether users could complete fundamental actions without difficulty.
Task 1
Open App
Task 2
Click the Sign Up Btn
Task 3
Fill in All the Blanks (about Your Personal Information)
Task 4
Read, Check the Terms of Service, and Finish Sign Up
Task 5
Swipe and Take a look at the characters
Task 6
Check the Character's information
Task 7
Choose the Character
Nothing failed functionally. What failed was role perception. Participants engaged the moment TRABU read as a companion, and pulled back whenever its character, its guidance or its use of their data stayed unclear.
Insight
Trust was decided before the product's value could be experienced. The idea of an AI travel companion was never the obstacle. The first minute was. So the design work moved to the earliest cues, what the character is, what it will do, and what it can see.
04 | outcome
solution
The solution had one job, to make TRABU legible before it is useful. A first-time user is told what the character is and what it can see, then given information they can act on where they stand. Play arrives after that, once the traveler already knows what they are traveling with.
" AR Travel Companion Experience "
Playful Travel Experience
Made solo travel feel more playful and immersive
Voice Interaction Flow
Made it easier to ask questions and follow spoken responses
Information Handoff
Smoothed the transition from spoken suggestions to external details
From Framework to Experience
The prototype puts AR navigation, voice guidance and character interaction on one continuous flow. It moves the traveler from orientation to discovery to play, in that order, because each step earns the next. Wayfinding is the floor here, not the goal. What the flow protects is the willingness to keep exploring.
Start
Personalized Setup
Onboarding sets travel style, preferences, and how much company the traveler actually wants. Asking that first is what makes the companion feel chosen rather than assigned. It is also where TRABU states what it will access, which is the exact step that lost people in the first round.
The prototype video may take some time to load. Please wait a moment.
Connect
Playful Character Chat
The first exchange is deliberately useless, small talk before any task. Meeting the character with nothing at stake is how a traveler learns its role without being taught it. By the time it matters, the voice is already familiar.
The prototype video may take some time to load. Please wait a moment.
Explore
Information Search & AR Navigation
The traveler asks out loud and gets a place, a reason, and an arrow pointing at it. Voice carries the answer while AR carries the direction, so neither one has to hold both. The recommendation stays checkable in one tap, because guidance that cannot be confirmed does not get followed.
The prototype video may take some time to load. Please wait a moment.
Play
Character Collection Game
Characters appear along the route, and only while the traveler is moving. Collecting them gives the walk between destinations a reason to exist. The reward is for exploring, so the game points back at the trip instead of into the screen.
The prototype video may take some time to load. Please wait a moment.
Watch Full Prototyping Video
design system
One visual language keeps the character, the AR layer and the interface reading as the same thing. Color, type and components are set against five qualities, openness, trust, curiosity, playfulness and immersion. Trust does the most work of the five, since it decides how playful the interface can be before it stops being taken seriously.
Openness
Trust
Curiosity
Playfulness
Immersion
Color System
Font System
Bold
34, 17px
ABCDEFGHIJKLMNOPQRSTUVWXYZ
abcdefghijklmnopqrstuvwxyz
1234567890’~-!?*
Regular
17px
ABCDEFGHIJKLMNOPQRSTUVWXYZ
abcdefghijklmnopqrstuvwxyz
1234567890’~-!?*
SemiBold
17, 13px
ABCDEFGHIJKLMNOPQRSTUVWXYZ
abcdefghijklmnopqrstuvwxyz
1234567890’~-!?*
OpenSans
UI Elements
05 | Validation
what i learned
"Translating Vision into Experience"
An idea about companionship sounds convincing until it has to become a flow, a distance, and a first minute. Testing showed the concept was never the thing people doubted. The interface was, before the concept could speak for itself. I learned to spend design effort at the entry, where trust is actually decided, instead of on the feature that makes the demo look good.
