We build apps

Siftware Dev, for the web

Siftware Dev is projects, time, scheduling, CRM and support, all in one app. We built it, we run Siftware on it, and every release goes through the same test suite before it goes near production.

It's our own, which is the point. We live with every decision we made in it.

Siftware Dev, for iPhone and iPad

A native app. Start a timer on your phone, walk into a meeting, and it's still running when you sit back down at your laptop. Lose signal on the train and you keep working: it syncs when you're back online.

Short lessons on growing food, for people learning on a cheap phone where data costs money. Lessons are audio and slides, saved to the phone to come back to.

Behind it is a backend we built for the job: members, invitations, lessons and sign-in, made to work over slow and patchy connections. Sign-in is a code sent by WhatsApp, through our own WhatsApp integration, so there's no password to forget and no email address needed.

Its website, permaculture.ac, is built the same way: one page that loads quickly on a cheap phone, with a sign-up form that works without JavaScript.

We've got more cooking, and we'll show them here once they're released.

Ever been handed an app that does exactly what the tickets said and nothing anybody needed? Us too. It's usually a brief problem, not a code problem.

So before we build, we write:

• What it's for. The job, in a page. Who uses it and what they do on a Tuesday afternoon.
• What "done" means. Agreed with you, not assumed by us.
• The design. Approved before screens are built, because moving a box in a mock-up costs minutes and moving it in code costs days.

Then we build, with tests running from the first commit. That's what lets us go quickly without it falling over the week after launch.

We've built our own apps with AI in the loop for two years. Used carefully, it makes a budget go further, and it's part of why a small team can build a whole system.

It doesn't replace an experienced developer. Every change starts with a test, and a person reads it before it ships.

How we build with AI

Every build comes with a plan for who runs it once it's live: us, your own team, or somebody else you choose. If it's somebody else, we hand it over properly, with the code, the docs and a working deploy.