Wedplan

One place for the whole wedding: checklist, budget, guest list with RSVP, seating plan, vendors, timeline, and a wedding website the couple builds themselves. Web, iPhone and Android, in sync, with scoped access for the people helping.

It is built for Switzerland rather than translated into it – Swiss francs, Swiss deadlines, German that sounds like German.

The build. React and TypeScript on the front, Supabase underneath – Postgres with row-level security, auth, storage, realtime and edge functions. The native apps are the same codebase through Capacitor rather than a second implementation, which is the only reason iOS and Android exist at all.

The parts that were actually hard are not the ones that look hard. The seating plan is an infinite canvas with resizable tables, drag-and-drop guest assignment and a PDF export – no library did this, so it is built. Access for a maid of honour or a parent needed scoped rights without an account and without touching the couple's own login, so it runs on hashed session tokens outside the normal auth path. And keeping three planners on the same numbers means live sync that survives a phone coming back from the background – a detail that decides whether a feature ships or quietly breaks on half the devices.

In two places, both narrow on purpose. There is an assistant inside the app that answers planning questions in context – it knows the couple's own budget and dates, so the answer is about their wedding rather than about weddings.

The bigger one is upstream: a product this size, on three platforms, does not get built by one developer in a few months without AI in the loop for the mechanical work. That is not a shortcut, it is the reason the scope was possible at all – and it is exactly the skill the other two projects also run on.