Thinking from Aqta

On building AI
for the real world.

Notes on AI in healthcare, agents in production, privacy as a design constraint, and the hard problems worth solving.

16 Jul 20265 min read

When routing is dynamic, "which model ran?" is an audit question

Teams are routing work across AI models per task to control spend, and the model serving a request is now decided at runtime. Good engineering, one new consequence: proving which model produced an output becomes a per-request question, and the answer should be a signed record.

EngineeringModel routing

12 Jul 20266 min read

A backtest is a claim. A receipt is evidence.

The largest banks are building AI agents for the investment process, and the evidence for what they can do is still the backtest. The stronger pattern: sign every agent decision before the outcome window closes, and the track record accrues as evidence rather than as a claim.

FinanceModel risk

29 Jun 20263 min read

The best argument against us

A regulatory advisor spent an hour trying to talk us out of building Aqta. Four questions that make the strongest case against verifiable AI receipts, and here are our honest answers.

PerspectiveAI accountability

31 May 20266 min read

The receipt problem with AI

Every other transaction in your life gives you a receipt. When an AI denies your claim, blocks your transaction, or recommends your treatment, what do you get? A plain-language explainer of the gap, with real cases and what a fix looks like.

ExplainerAI accountability

17 May 20268 min read

Eight days before the WHO. AqtaBio's first prospective Ebola signal, and what it does and doesn't prove.

On 9 May 2026 the inaugural entry in our append-only public ledger listed all five top Ebola tiles in the Congo Basin biome. Eight days later the WHO declared a PHEIC for Bundibugyo Ebola in DR Congo and Uganda. Biome-correct; the country was rank 4, not rank 1. Commit #1, not #100. First prospective signal. The lead is real, the caveats matter, the verification is non-custodial.

AqtaBioPre-etiologic surveillance

8 May 20268 min read

Who's accountable when healthcare AI makes a mistake?

Ireland's Medical Council says doctors remain responsible for AI decisions. But how can they be confident in tools they don't fully understand?

HealthcareMedical AI

1 May 20266 min read

The human supply chain behind AI

Three AI coding assistants, same task. All three imported deprecated libraries. One pulled a known RCE flaw. None flagged it. The model is the easy part.

Supply ChainEthics

26 Apr 202611 min read

How AqtaCore signs attestation receipts: Ed25519, canonical JSON, and zero-trust verification

A walk through the cryptography behind every AqtaCore enforcement decision: Ed25519 signing, the canonical-JSON rule that keeps Python and TypeScript verifiers byte-identical, and how an auditor verifies a receipt in five lines without ever calling our servers.

AqtaCoreEd25519