The meter changed how people build.

Watch someone use a metered AI builder for the first time. They describe what they want. It appears. They laugh — genuinely, out loud. It's one of the few moments in software that still does that to people.

Now watch them a fortnight later. They write the prompt. They read it back. They edit it. They think about whether this is worth an attempt. Something in the room has changed, and it isn't their skill.

Here is the mechanic underneath. The model writes code. Some of that code is wrong — not occasionally, routinely, that's what the technology is. You then spend attempts getting it right. Every one of those attempts is billed to you. So the worse the model performs, the more revenue it generates, and the person paying for the mistake is the person who didn't make it.

Nobody sat in a room and designed that as a punishment. It's just what happens when you price by consumption and the consumption is largely the vendor's own error rate. But the effect is precise and it's felt by everyone: the tool that was supposed to make you brave made you cautious.

So we took the meter out.

Not hidden, not raised, not renamed as "credits" or "tokens" or "flex units". Removed. You pay a flat monthly price. You iterate as much as you like. The month you spend hunting a bug the model wrote costs the same as the month everything went right, which is the entire point, and is the only claim on this site we would fight about.

This is priced like insurance rather than like electricity. Most months you'll use less than you pay for. Those months pay for the ones where you hammer it for three weeks straight. We took the risk of a bad month off your desk and put it on ours, because we are in a much better position to absorb it than you are.

Then we fixed the other three fears.

The morning betrayal. The app worked last night and doesn't now, and nobody can say what changed. Every version is kept, the database travels back in time with the code, and a restore opens a read-only copy alongside the live one rather than overwriting it. You take back what you lost. Nothing goes offline while you do.

The leak. Secure defaults, keys that never reach the browser, tables private until you say otherwise, and a security report you can actually read. We will not tell you your app is unhackable. Nobody can honestly tell you that. We'll tell you what we checked, what passed, and what we didn't check.

The hostage. Your code exports, in a standard stack, on every plan including the free one. There is a door and it is always unlocked. A product that has to lock you in has admitted something about itself.

What we're not going to do.

We're not going to invent a founder story with a garage in it. We're not going to put five-star ratings on this page before anyone has given us one. We're not going to run a countdown timer on a price that will still be there next month.

We don't have testimonials yet, because we don't have enough customers yet to have honest ones. When we do, they'll be real, attributed, and linked to a person you can go and ask. Until then this space stays empty. An empty space is more useful to you than a fabricated one.