I've been building mobile apps for ten years. Joru is the one I wanted to exist.
I've spent the last ten years developing mobile applications. That's long enough to form some firm opinions about what makes an app worth keeping on your phone: it should do one thing properly, respect the device it runs on, and never ask for more than it actually needs.
Joru is built and maintained by me alone — the app, the data pipeline behind its offline place extracts, and this website.
I've followed AI closely for years: not just how to call a model, but what running one really costs, where it fails, and what it should never be trusted with. That last part matters most here — a language model is excellent at arranging and describing, and genuinely unreliable at remembering whether a restaurant exists.
When models small enough to run on a phone became good enough to do real work, running the AI on-device stopped being a curiosity and became the obvious way to build this app.
I like travelling, and I've planned plenty of trips with the tools that were available. Looking hard at what itinerary apps let you do — and precisely where they stop — is what pushed me to build Joru: too many limits on how much you can plan, too much personal data leaving the device, and too much invention presented as a recommendation.
So Joru is built the other way round. The facts come from open data before the AI is involved, the numbers are computed rather than guessed, and the model runs where your data already is: on your phone.
