A GPS audio walk through Willemstad, from the first workshop to the field test.
Tourists in Willemstad fear wasting their few hours on a poor tour. We built one walk they can trust, and the way to prove it works.
A wide tourism idea, no proof anyone wanted it, and little time on the island.
Led the MVP, the design system and the full build in code. Shaped the test plan.
A working GPS audio walk, load-tested and secured, ready for the field test. Live at dowigo.com and app.dowigo.com.
Thick line: I led it. Thin line: I contributed to work someone else led.
Navigation benchmark, test plan, success measures.
Walker journey, core flow, thirty stories.
The clickable walk, and every follow-up.
The system the app is designed with.
Walker app, admin, analytics, in Claude Code.
What to measure, consent, the sprint write-up.
The rule on the day two slide: we do not design yet. We decide what deserves to be built first.
Reach the start with confidence.
Open the app, read the story disclaimerStart without any friction.
Tap Start tour, see the way to stop oneConnect with the place and its history.
Arrive, and the story playsNever feel lost between stories.
Follow directions to the next stopClose the walk and say how it was.
End the tour, rate itAs a tourist mid tour, I want to see which stops I have done, so that I see my progress.
As a tourist, I want the path to the next stop once a story ends, so that I never feel lost.
As a tourist with little time, I want a trailer of the tour, so that I can decide if it is worth it.
As a tourist with limited hearing, I want captions, so that I can follow the story.
As an admin, I want a test tour in Amsterdam, so that we can test without being in Curaçao.
As a walker who hits a bug, I want a clear way to recover, so that I can carry on.
As a tourist, I want to know I completed something meaningful, so that I feel fulfilled.
As a tourist, I do not want money decisions during the walk, so that I am not distracted.
Three calls from the workshops. Scroll to see the first one happen.
Navigation and stories are worth nothing apart, so the MVP is those two and nothing else.
1curated walk
What it cost: no video, no AR, no tours made by users, and English only at first.
Success was defined before the first screen, so analytics and a feedback screen are part of the MVP.
What it cost: sharing, likes and tipping waited.
STOP 1 OF 8Start @ Curaçao SignThe pattern from hiking and map apps, and only what a story walk needs. No fixed start: you are routed to the nearest stop.
What it cost: no spoken turn-by-turn. Walkers read the route.
The Figma Make MVP proved the walk in March. From May I rebuilt it as a full product in code: walker app, sign-in for paid tours, admin and analytics.
Two questions for the island: can people walk on their own, and does the app hold up when many use it at once?
14 of 23 "completed" walks had not visited a single stop. Now the count, and the walker's own screen, follow what really happened.
One day of work instead of four. Each use is logged, so we still see how often GPS misses in narrow streets.
The field test was planned for late September 2026 and has moved. Its results will land here.
What real walkers do. And whether audio-walk people are the guided-tour people.
Define what counts as a finished walk before building the thing that counts it.
Next case studyAI film pipelineHow a studio makes AI films that look authored, on any model.