For most of the past decade, maritime risk products have competed on the cleverness of their score. A vessel was assigned a number — a colour, a tier, a probability — and the customer was expected to trust it. The methodology was a competitive secret. The output was a verdict, not an argument.

That worked, briefly, when the alternative was nothing. It does not work now.

What changed.

Three things, in roughly this order. First, regulators began to ask insurers and trading houses to defend their decisions, not just make them. A red flag from a vendor is no longer a sufficient basis for declining to underwrite. Second, the trades that matter — sanctioned crude, dual-use shipments, evasion through opaque chartering — became sophisticated enough that simple identity-based screening misses them. Third, customers got tired of paying for opinions they could not audit.

Black-box systems tell you what to think. Glass-box systems show you what happened.

What we mean by glass-box.

A glass-box signal carries three things along with it: the observation that triggered it, the baseline it was compared against, and a confidence interval that distinguishes confirmed evidence from probable inference from open question.

Concretely: if HUAX raises an STS proximity signal, the customer can see the AIS fixes that produced it, the peer-group baseline for that lane and that season, and an explicit statement of which parts of the chain are direct observation and which are inference. If the inference is wrong, the customer can challenge it. If the observation is wrong, the customer can correct it. The system does not pretend to be infallible — it pretends to be readable.

Why now.

HUAX was founded in 2024 because a small set of buyers — insurers, brokers, compliance teams, investigators — were asking for the same thing in different words. They wanted intelligence they could put in front of a regulator, a board, or a court. They wanted to be able to say, in plain English, why the system raised a signal. They wanted to be able to disagree with the system, and have that disagreement be useful.

That is what we are building. Not a faster red flag. A defensible one.