Why I built PlanItShare

I travel more than is probably reasonable, and for years I've relied on TripIt to keep flights, hotels, and half-remembered plans in one place. It's good software. But every trip also comes with a second, messier spreadsheet: who paid for what. My friends and I settle up trip expenses — and honestly a lot of our regular group hangouts too — through Revolut, which handles the money fine but has never once cared what the money was actually for. TripIt won't split a taxi fare three ways, and Revolut doesn't know you went to Chalkidiki in August. Somewhere between those two apps there was a gap exactly the shape of a weekend project. That, and I wanted a real reason to build something end-to-end with Fable 5 instead of just reading about it. PlanItShare is what happens when "I should try this new model" collides with "why do I still do this math in my head" — and it's open to my friends, and honestly anyone else who wants in.

Next.js on the frontend, Supabase for the database, auth, and storage, Tailwind for styling, Mapbox for the maps, and Claude doing the unglamorous work of reading your forwarded flight confirmation emails and figuring out what's actually in them. Nothing exotic — the interesting part isn't the stack, it's how it got written.

Not with one AI tool — with a small assembly line of them. Claude Code did essentially all of the design and implementation, from the database schema to the button you're about to click. Perplexity handled the deep research and helped me write the instructions I fed to Claude, which — if you've ever tried to brief a very capable but very literal junior engineer — is its own skill. Grok handled some of the imagery. No single tool gets full credit, and none of them get none of it.