Case Study

Bon Voyage: how long can I stay?

Bon Voyage tells Canadians how long they can stay in the US without breaking the 182 day rule. I designed it, then built and shipped it myself with Claude Code. It’s live as a web app, and a frequent cross border traveler uses it to plan real trips.

ROLE

Research, product design, front end build, deployment

USERS

Canadians who fly to the US often. I worked closest with one, and his Canadian friends had the same problem.

TIMELINE

September-November 2026

TOOLS

Claude, Claude Code, Vite + TypeScript, Vitest, GitHub, Vercel

The problem

Canadians can spend at most 182 days in the US in any rolling 12 months. Not a calendar year, so there’s no reset date. The people I talked to were tracking it by digging up old flight confirmations and doing the math by hand.


  • The rule is hard to hold in your head. Days come back one at a time, each exactly a year after you spent it. Between those anniversaries the count sits flat, sometimes for weeks.

  • Even frequent travelers get it wrong. My main user counted his trips in nights and got 162. The rule counts days, so the day you land and the day you leave both count. The real number was 167.

  • Existing apps miss the mark. I found at least 8 (Snowbirds US Day Tracker, SnoBird, Flamingo Compliance, TrackingDays and others) at $4 to $10 a month. They’re built for retirees, and reviews complain about backfilling old trips by hand, flaky GPS tracking and dated UI.

The opening: a younger cross border traveler who wants one clear answer, fast backfill, and something that doesn’t look like tax software.

Key decisions

01

Lead with one answer

Every tracker I looked at opened with a trip log and a running total. That’s backwards. Travelers don't care how many days they have used. They want to know when they can go and how long they can stay.

So Home is one card. Pick a fly in date, and the biggest thing on screen is how many days you can stay and the date you have to leave. Arrows move a day, chips jump a week or a month, and a calendar reaches two years out. If the date lands inside a planned trip, the card says so and links straight to it.

02

Show the number he can defend at the border

The math has a twist. On a long stay, old days drop out of the window as you go, so the true limit can run longer than 182 minus what you’ve used. My first version showed that longer number. Beau saw 182 when he expected 172 and figured it was broken.

So I looked at how the rule actually gets enforced. The 182 days isn’t a hard count written into law. It’s how officers judge whether you’re visiting or living here, and they look back 12 months from the day you arrive. Now the big number is the safe stay, the one an officer would land on. The longer number lives behind a setting, and buffer days let him keep a margin.

03

Cut anything that needs explaining

I designed a strip chart of days available by fly in date. It was accurate and almost unreadable. The bars sat flat for weeks, then jumped, and every jump needed a paragraph to explain. Same with a summary card at the end of onboarding, and a term I’d made up, “new US days.”

If a component needs a footnote to justify itself, it goes. The strip, the card and the term are gone, and Home fits on one screen.

04

Make the backup something a person can read

A JSON file is fine for a machine. One user I spoke with keeps his life in Notes. So exports are a plain list: one trip per line, the year on every date, the route, and whether it’s planned. The same list pastes straight back into the app with nothing lost, and pasting it twice doesn’t double anything.

Results

  • Built and shipped a live web app with Claude Code. A frequent cross border traveler uses it to plan real trips.

  • Installed on a real user’s home screen with his trip history loaded.

  • This was my first time owning the dev side of something end to end: repo, tests, deploys, and keeping a real person’s data safe.