Field notes on building software and AI agents for real businesses.
The default for most AI tools is: send your data to their servers and trust their promises. For a small business handing over its operations, that is a hard sell, and it should be.
So ALEA builds the production app to run inside your own cloud. The agent works on your data where it already lives, and nothing gets shipped off to be trained on. Sovereignty is not a feature bolted on later; it is the shape of the thing from the start.
The studio builds a working prototype in minutes, free. That is not generosity. It is the fastest way to agree on what we are actually building before anyone spends real money.
The prototype settles the argument: here are the screens, here is the workflow, here is the agent answering a real question. Once that is right, the production system gets built by hand, deployed, and handed over. The demo is the conversation; the build is the deliverable.
A timesheet used by crews across mine sites has no room for cleverness. It has to work with gloves on, on a bad connection, at the end of a long shift, for people who never asked for an app.
That constraint is a gift. It strips a tool down to what matters: the fewest taps, the clearest state, nothing that breaks when the network does. Software that survives the field respects the person using it, which turns out to be true everywhere, not just underground.