Every venue now claims AI. Almost none of those claims survive a conversation with a model risk team, because the intelligence was bolted on where governance cannot see it. We inverted the build order and drew the perimeter first.
We drew the perimeter before we built the mind.
That sentence is the entire design philosophy of AIDANN, and what follows is what it means in practice.
Every venue now claims AI. Almost none of those claims survive a conversation with a model risk team, because the intelligence was bolted on where governance cannot see it.
The barrier to adoption in institutional markets has never been capability. It is trust, and trust in this world is not a feeling. It is an audit trail.
So we inverted the build order. AIDANN, our intelligence layer, stands for AI for Data-Driven Analytics, Navigation and Netting, and it is being built to operate across four domains: liquidity sensing, execution guidance, settlement intelligence, and limit awareness. But the architecture was drawn before the intelligence was. A governance perimeter defines exactly what AIDANN can see, what it can say, and what it can never do.
Here is the part that matters most, and it is one sentence.
AIDANN holds no execution permissions. It cannot place an order, amend an order, or cancel an order. Not because a restriction was bolted onto a powerful system, but because the permission does not exist in the architecture for it to hold.
AIDANN describes outcomes it cannot touch.
Every figure AIDANN reports comes from the deterministic engine, delivered through a fixed template with computed slots. Open the template and you can read exactly which words are fixed and which numbers were computed.
No paraphrase. No improvisation. No generated value anywhere in the chain.
There is a simple test for whether intelligence is an amplifier or a dependency, and we build to pass it.
If AIDANN were switched off tomorrow, the netting would still run. Participants would read a table instead of a paragraph.
The engine does not need the intelligence to be correct. The intelligence makes the correctness legible, navigable and useful at speed. That is what governed intelligence means, and it is the only version of AI we believe institutional markets will ever run at full capacity.
The next decade of trading will not be won by the venue with the most intelligence. It will be won by the venue whose intelligence your governance can trust at full speed.
A brilliant system your compliance team meets for the first time in an incident report is not an asset. It is a liability with good marketing.
Open the engine, let it run, and watch the describe cycle: the engine computes, then AIDANN speaks, in that order, every tick.
Then open the template and try to find a single value that did not come from the engine. That challenge stands permanently.
If you think the perimeter should be tighter, say where. That is a genuine question, and the sharpest answers so far have come from people who deal with model risk for a living.