The best argument against us

Founder's note

A regulatory advisor I respect spent the better part of an hour trying to talk me out of building Aqta. He had four questions. They were good ones. Here are my honest answers.

“Isn't this just a feature? Any gateway can sign a log”

He's right that signing a log is a feature. Any proxy can append a hash and call it a receipt. But a signature you have to trust the signer to believe isn't evidence, it's a claim with extra steps. The company was never the signature. The company is being the standard anyone can verify without trusting us, across every model and vendor. Stripe was “just a feature” on top of the card networks too. The feature was never the point. The position was.

“If you sign your own receipts, why would anyone trust them?”

They shouldn't. That's the design. An Aqta receipt is Ed25519-signed and verifiable offline, with our open-source verifier, without ever calling our servers. If Aqta disappeared tomorrow, every receipt we ever issued still verifies. You aren't trusting us that a decision happened a certain way. You're checking the math yourself. To be exact about what that buys you: it is tamper-evident proof of what was claimed and that nobody altered it after the fact. It is evidence, not a court's verdict. But “here is a record you can verify without me” is a very different conversation from “trust our dashboard.”

“How many companies actually run automated decisions that get challenged?”

More than you'd think, and rising fast. But the raw count is the wrong number to argue about. The one that matters is how many of those decisions get contested, and what happens on that day. A rejected job applicant now has case law behind them (Mobley v. Workday).^[1] A denied claim, a blocked transaction, a declined loan. On the day one of those is challenged, “the model decided, and we logged it somewhere” is not an answer a regulator, a court, or the other side's lawyer accepts.

So when does this actually matter?

Not for the routine ninety-nine percent of decisions. For the contested one percent. Independence is worth nothing until a decision is questioned, and close to everything the moment it is. We aren't overhead on every call. We're the receipt you're very glad you kept on the one that goes to a hearing.

He didn't fully agree by the end. I didn't need him to. The point of building in the open is that the strongest objections sharpen the thing instead of sinking it. If you've got a better argument against us, I genuinely want to hear it.

And if you are weighing whether this matters for your use case, that is exactly what our pilots are for.