Md Aseer Intiser · aseer.design/work/travelmate-aimd.aseerintiser@gmail.com
TravelMate
A travel companion that suggests what to do right now, explains why, and rewards every real place you visit with an ink stamp. A self-directed concept, redesigned from the ground up in 2026.
- Role
- Product Designer (solo)
- Category
- Mobile Design, Product Design, AI UX
- Duration
- Jan – May 2025, redesigned Oct 2026
- Tools
- Figma, Claude (AI pair designer)
- Status
- Self-directed concept, not yet user-tested
- Reading time
- 5 min
In short · 5 min read
- The problem
- Travellers juggle maps, translators, review sites and group chats, and most travel apps reward planning at home rather than being out in the city.
- My role
- Sole product designer on a self-directed concept, with Claude as an AI pair designer for producing the Figma file.
- Key decision
- Make the reward travel-native: real places earn ink stamps and reveal the map, and every AI suggestion shows its reason and how sure it is.
- Result
- A 17-screen iOS concept with a token-based design system in Light and Dark mode, and a 40-link prototype.SourceWCAG contrast audit of the final palette, measured in the TravelMate Figma file, October 2026.
Chapter 01
The short version
TravelMate began in 2025 as a chat-first travel assistant: ask a question, get a map. It worked, but it looked and behaved like every other AI travel concept. In October 2026 I rebuilt it from the product idea up.
The new version has one idea at its centre. Most travel apps reward you for planning at home. TravelMate rewards you for going. Tuki, a small travel guide, suggests something good nearby and explains why. You go, you check in, and you earn an ink stamp, and that part of the city fills in on your map.
This is a self-directed concept, not a shipped product, and nobody has tested it yet. I say that up front, because the rest of this page is about decisions, not results.
- Screens, in journey order
- 17
- Linked prototype interactions
- 40
- Modes, Light and Dark, from one set of tokens
- 2
- Contrast failures found and fixed
- 6
Chapter 02
The idea: reward going, not planning
I started from the moments that matter on a trip: lost, hungry, alone, unsure. In every one of them, the traveller is out in the city, not at home planning. So the product should reward being out there.
Turn a trip into something you collect.
The game layer is travel-native on purpose. Early on I tried a game-style direction with a lesson path and chunky 3D buttons, borrowed from a language-learning app. It taught me the most, because copying the surface of a successful app is not the same as learning from it. So the rewards here come from travel itself. A stamp only comes from being somewhere real, checked within 40 metres of the place, and there are no streaks to protect, because a travel app that punishes a rest day has misunderstood travel.
Chapter 03
Now: a map, one suggestion, and the reason
Travellers think in places, so the home screen is a map. Areas you have explored are clear, and the rest is softly hatched, like a scratch map. Over it sits one suggestion, not a feed.
Close-up · the Now screen
One main action per screen, in ink black, and help one tap away. Everything else on the map is there to answer one question: is this worth walking to right now?
Chapter 04
Designing for trust
This is the chapter I care most about. My master's thesis studied how international students come to trust a conversational AI, and what breaks that trust. One finding carried straight into TravelMate: trust builds slowly and can vanish in a single moment. So Tuki explains itself every time, and handles its own mistakes in the open. Tuki has four moods, and the newest one, Sorry, exists for exactly that moment.
Four trust moments
The reason, the sources, and how sure Tuki is, in plain words.
Confidence is shown as "fairly sure" on a three-step meter, never as a precise-looking percentage. A number like 82% would claim an accuracy the app cannot have.
Chapter 05
The reward: a stamp, and a city that fills in
Check-in only works within 40 metres of the place. The stamp lands with a short drop and a small ink burst, and the neighbourhood's share of the map goes up. The Passport keeps every stamp from the trip, plus a plain list of how much of each area you have revealed.
Chapter 06
Everything else a trip needs
The loop is the heart of the app, but a trip needs more. Each of these screens answers one moment from the journey map: arriving, ordering, asking, remembering, meeting people, and things going wrong.
The rest of the trip
One promise and one button. No sign-up wall: you only need an account to keep your stamps.
Help now is the one screen without Tuki. When something goes wrong, a cartoon guide is the last thing you want to see, so that screen is calm, fast, and works without signal.
Chapter 07
A system, not a set of screens
Every colour comes from 17 semantic tokens, each with a Light and a Dark value. JetBrains Mono carries ticket-style data like times and percentages, so data reads differently from conversation.
Same screen, two modes, nothing redrawn
The Dark version is the same screen with one setting changed. Building it was also a test: it exposed two contrast failures that only appeared in Dark mode.
Chapter 08
Accessibility, measured
A travel app is used outdoors, in a hurry, and sometimes in an emergency. So I measured instead of guessing.
The first audit found six pairs below WCAG AA. Five were fixed by changing a token, so every screen that used them updated at once. The sixth, the faded map labels, was a hard-coded colour that I fixed by hand.
- Touch targets are at least 44 by 44 points, with small icon buttons padded to reach it.
- Colour is never the only signal. Explored areas also get a tick and a percentage.
Chapter 09
What this does not prove yet
Nobody outside this project has used TravelMate yet, so none of it is validated. Two assumptions carry the whole concept, and they are what I would test first, with five travellers and the prototype. Do stamps motivate adults to explore, or do they feel like a gimmick? And does showing the reason make people trust a suggestion more?
I used Claude as an AI pair designer while producing the Figma file. The product idea, the decisions and the trade-offs on this page are mine, including the directions I threw away.
What I am confident about: the core idea is specific, the trust patterns come from real research rather than taste, and the system holds together across seventeen screens and two modes.