How capital reads your AI, made knowable in advance.
VeriQ reads a company AI the way a serious investor does, on evidence, against a single readiness instrument. The lens is knowable in advance, which means the gaps are findable in advance, and closeable before the meeting.
What AIQ is, and is not
AIQ is a readiness instrument. It is scored 0 to 100, rendered headline first, for example "AIQ 78/100" with a one-line verdict, and it is shown privately only. It measures the evidence behind an AI claim, the way an institutional diligence team would.
- Fees never buy a score. The methodology is the methodology, and the evidence either exists or it does not.
- Scores are private. No client is named, and below-the-line results are never displayed.
- Assessment and advisory are separate and disclosed. Finding a gap and building past it are different engagements.
- Independence is the product. A read you cannot trust is not worth running.
The questions a diligence team works through.
None of this is exotic. It is the standard global capital now applies, and the uncomfortable part is that most teams, asked these questions plainly, can evidence fewer than half.
Provenance and IP
Who built the model, on whose data, under what licence, and what survives if a key engineer or vendor leaves. A strong answer names the sources, the rights attached to each, and the line between what you own and what you rent.
Human oversight
Where a person sits in the loop, who can intervene, and what is logged when the model acts. Weak oversight reads as unpriced risk. Clear oversight, with named owners and a record, reads as a team that understands what it has deployed.
Failure modes
What happens when the model is wrong, because every model is wrong sometimes. The teams that pass can describe how often, in what direction, at what cost, and what catches the error before a customer or a regulator does.
Unit economics of the AI claim
What the AI costs to run per unit, per customer, per inference, and whether that cost falls or rises as you scale. A value story that ignores the cost of compute, data, and oversight re-prices the moment someone does the arithmetic.
Vendor dependency
How much of the capability rests on a single provider, API, or contract, and your position if that provider changes terms, price, or model overnight. A robust answer maps the dependency and shows a path if it moves.
Attribution of uplift
Of the improvement you attribute to AI, how much is the model and how much is everything else you changed at the same time. If you cannot separate the two, the room assumes the smaller number.
Aligned to the field, endorsed by no one.
VeriQ aligns its questions to the public bodies of work converging on this discipline, including the Institute for Ethics in AI, and frameworks such as the NIST AI Risk Management Framework and ISO 42001. These are referenced as external work the instrument aligns to. VeriQ is not affiliated with, endorsed by, or approved by any of them, claims no recognition or certification, and uses no logos.
Free download, the Diligence Probe Library
The real evidence questions a serious investor works through when AI is load-bearing. Score your own answers in private. Marked indicative, unaudited, public information only.
Download the Diligence Probe Library