VeriQ
Independent · Private · Evidence-based · Hong Kong
AI readiness, scored the way global capital scores it

Build your AI to the standard investors recognise, before they ask.

VeriQ Standards is an independent, Hong Kong based readiness methodology. Fifteen years inside global investing, distilled into the 30 to 40 evidence questions diligence now asks about your AI, and a clear path to answer every one.

When the diligence call turns to AI, most teams can answer fewer than half the questions. That gap is fixable, before you are in the room.

One question worth answering first. When does your AI claim go to print, or when is your next LP or investor AI-diligence cycle? If it is inside ninety days, this is the right moment.

What you actually receive

A readiness read, headline first.

Every read leads with a private AIQ out of 100 and a one line verdict, then the specific places a diligence room will press, and the exhibits that close each gap. Here is a composite, so you can see the shape of it.

Anonymised composite. Describes no real firm. Any real AIQ score is private.

70/100 Fundable once the evidence behind two claims is assembled.
0observed ceiling 86100
Measurement and banked gains

The headline lift is asserted, not shown. There is no held-out cohort, so the number cannot survive a second question.

Data and IP defensibility

The model is real, but the right to the data it trains on is undocumented, which is the line a diligence team draws conservatively.

See the full sample read, all six sections →


Start free, with the 5-question self-check

Five of the questions an institutional investor's diligence will ask. Score yourself in two minutes. Private. No score is ever published.

You also receive the Diligence Probe Library, the real evidence questions a serious investor asks, marked indicative and public information only. No spam, unsubscribe anytime. See the method →
The five questions your investor will ask
  • Provenance and IP. Can you show what data trains your top AI use-case, who owns it, and the licence you use it under?
  • Human oversight. Where AI shapes a decision, can you show the documented control, and the last time it caught an error?
  • Failure modes. Is there a written list of how your AI can fail, the monitoring that catches each, and a named owner?
  • Unit economics of the AI claim. Can you separate the revenue and cost your AI actually drives, with assumptions written down?
  • Vendor dependency. If your key model vendor changed terms or price tomorrow, what is your documented fallback?
Why a build, not a slide

You walk in with the evidence already assembled.

01

The dream outcome

Your AI claims hold under evidence questioning, because the evidence was built before you sent the deck. The score stays private to you.

02

Ten working days

From kickoff to a complete read. Not a quarter. You are usually building against a clock, so the work moves at that pace.

03

Effort, ours not yours

You sit the assessment and hand over what already exists. We construct the evidence pack, write the remediation, and rehearse the answers.