If an auditor showed up in90 days, would you pass?
Find out before someone asks. This triage scores your AI program across seven control domains and gives you a straight answer — AUDIT-READY AS DESCRIBED, PATCHABLE, or WOULD NOT PASS — plus, where anything is short, the one thing to fix first. With a gate that fails any in-scope control nobody owns, whatever the score.
A readiness self-assessment and decision aid — not legal advice and not an audit opinion. It does not certify compliance. Confirm scope and evidence with your own counsel or auditor.
Find your gaps before the auditor does.
Of senior US business leaders lack full confidence their organization could pass an independent AI governance audit within 90 days, per Grant Thornton's 2026 AI Impact Survey of nearly 1,000 executives.
The question this triage asks before it reads the score. A domain with great documentation and no accountable owner fails on the spot.
The window the Grant Thornton survey asked about, and the one this triage is built around. A real request sets its own deadline; know your answer before the clock starts.
Score the domains. Get the verdict.
No-owner gate: 1 in-scope domain has no named owner. That fails the audit regardless of the score.
Vendor & sub-processor governance — no named owner
Tap an evidence cell (0–3), flip an owner, or drop a domain out of scope and watch the verdict move. Worked example as of 2026-06-25.
| Control domain | Wt | Evidence | Owner | In scope |
|---|---|---|---|---|
| AI system & use-case inventory | 18 | |||
| Named owner & accountability | 14 | |||
| Acceptable-use policy & approvals | 14 | |||
| Output record-keeping & audit trail | 16 | |||
| Vendor & sub-processor governance | 14 | |||
| Risk assessment & monitoring | 14 | |||
| Staff AI-literacy / training record | 10 |
Same math as the workbook: weighted evidence renormalized over the in-scope domains, and a no-owner gate that forces WOULD NOT PASS when any in-scope control lacks a named owner — the way the shipped example scores 92 yet still fails. Fix-first names the heaviest ownerless domain while the gate is firing, otherwise the weakest evidence, and names nothing once every in-scope domain is owned and at 3. A readiness self-assessment, not legal advice or an audit opinion.


How the gate works, in one image
The seven weighted controls, the band thresholds, and the no-owner gate that forces WOULD NOT PASS — the same logic the demo runs, as a diagram you can share or embed anywhere.
View & embed the full diagramA program at 92. One uncomfortable result.
This is the example the workbook ships with, as its own formulas score it; the demo above opens on the same marks. It scores 92 out of 100 — comfortably above the AUDIT-READY AS DESCRIBED floor — and still lands WOULD NOT PASS, because one control domain has no named owner. The score alone would have waved it through. The gate didn't.
"Pass a 90-Day AI Audit?": Readiness Triage tab, as the workbook ships =========================================================================================== Control domain Weight Evidence (0-3) Owner (1/0) In scope (1/0) AI system & use-case inventory 18 3 1 1 Named owner & accountability 14 3 1 1 Acceptable-use policy & approvals 14 3 1 1 Output record-keeping & audit trail 16 3 1 1 Vendor & sub-processor governance 14 2 0 1 Risk assessment & monitoring 14 3 1 1 Staff AI-literacy / training record 10 2 1 1 ------------------------------------------------------------------------------------------- Readiness score (0-100) 92 Ownerless in-scope domains 1 VERDICT WOULD NOT PASS Fix first Vendor & sub-processor governance
Three rules keep the verdict honest.
Any in-scope domain with no named owner forces WOULD NOT PASS — full stop. A control nobody owns has no one to attest to it, so ownership gates the verdict.
Mark a domain out of scope when it genuinely doesn't apply; the weights renormalize over what's left. No padding, no distortion.
Where anything is still short, the verdict names the single highest-leverage gap — the ownerless control if the gate fired, else the weakest evidence. Once every in-scope domain is owned and at 3 it names nothing.
A readiness read, not a compliance certificate.
- A cross-domain triage of whether you'd pass a 90-day audit.
- An accountability gate that catches ownerless controls.
- A remediation runbook that sequences the fixes — owners first.
- An audit opinion or a compliance certification.
- Tied to one framework — it triages the domains they share.
- Legal advice. It tells you where you stand; your counsel rules on sufficiency.
A readiness self-assessment and decision aid — not legal advice and not an audit opinion. It does not certify compliance. Confirm scope and evidence with your own counsel or auditor.
Whoever has to answer when the request lands.
Triage first, then go deep where you're weak.
Go deep on the record-keeping domain: who generated what, reviewed by whom.
ViewClose the vendor-governance gap with a defensible data-flow register.
ViewBuild the full framework once the triage shows you're serious.
ViewEverything else you'd ask before buying.
Any review where someone asks for evidence that your AI is governed — a customer's vendor-risk or security review, a SOC 2 / ISO readiness check, an insurer's questionnaire, or a regulator's request. The triage isn't tied to one framework; it scores the seven control domains those reviews share — inventory, ownership, policy, record-keeping, vendor governance, risk, and training — so you get one cross-domain read on whether you'd pass.
Because accountability is a gate, not a line item. Any in-scope control domain with no named owner forces a WOULD NOT PASS verdict regardless of the score — the triage asks “who owns this?” before it reads the score, because a control nobody owns has no one to attest to it, maintain it, or answer for it. In the shipped example, the program scores 92 but its vendor-governance domain has no owner, so it fails. A strong score can't paper over a control nobody is accountable for.
AUDIT-READY AS DESCRIBED: every in-scope domain is owned and your weighted readiness is 80 or above — you'd likely pass. PATCHABLE: everything is owned but the evidence is thin (readiness 55–79) — fixable inside the 90-day window. WOULD NOT PASS: either a domain has no owner (the gate fired) or readiness is below 55. Where anything is still short, the verdict also names the single domain to fix first — the ownerless control if the gate fired, otherwise the weakest evidence. Once every in-scope domain is owned and at 3 it names nothing, because there is nothing left to repair.
This sits above them. The AI Output Audit-Trail Kit goes deep on record-keeping, the Vendor & Sub-Processor Data-Flow Register goes deep on vendor governance, the AI Literacy kit evidences training, and the NIST AI RMF kit builds the full framework. This triage is the cross-domain rollup that tells you which of those domains is your weakest — the front-door “which gap do I close first” read that routes you into the right deep kit. Run it first, then go deep where you're weak.
Yes. Flag a domain out of scope when it genuinely doesn't apply to you, and the weights renormalize over what's left — no padding, no distortion. One guardrail: an out-of-scope domain can't fire the no-owner gate, so dropping a domain just to dodge the accountability check won't help. Scope it honestly and the verdict stays meaningful.
No. It's a readiness self-assessment and decision aid — not legal advice, an audit opinion, or a compliance certification, and it doesn't certify anything. It tells you where you'd stand and where to focus; your own counsel or auditor rules on whether the evidence is sufficient. The download is one .xlsx workbook plus two .docx guides, a Facilitator Playbook and a 90-Day Remediation Runbook; the workbook is deterministic and works offline.
Know your answer
before they ask.
One purchase, lifetime access, 12 months of updates. $79, once.
A readiness self-assessment and decision aid — not legal advice and not an audit opinion. It does not certify compliance. Confirm scope and evidence with your own counsel or auditor.
Sold by RedHub AI LLC · Secured by Stripe · redhub.ai