Skip to content
Graziul Advisory

Free. Instant. Useful even if you never hire me.

The AI Governance Durability Check™

If you never pay me a dollar, this still gives you a read you didn't have and couldn't easily generate yourself. That's the commitment.

The check estimates how long your current AI governance will stay valid before it breaks — structured around the eight structural causes that documented governance failures keep tracing back to. Fifteen questions, about ten minutes, scored the moment you finish.

You get a durability band — Brittle, Aging, or Durable-ish — with the specific structural gaps that put you there, not a vanity number to optimize: the three stalest parts of your governance and the failure each one mirrors, the trigger to worry about first, and a three-step 90-day plan. Every part of the read traces to an answer you gave — nothing is a black box. It runs entirely in your browser; your answers are not sent anywhere unless you choose to email them.

About you (used only to segment your read)

Company size, sector, and the jurisdictions you operate in.

Used only to segment your report — not scored.

Which Gen-AI tools are in active use?

Multi-select. Used only to segment your report — not scored.

Policy vintage & ownership

When was your AI policy last substantively revised?

Who owns AI governance?

Does your AI policy name specific vendors, model versions, or capability thresholds?

Naming hard-coded specifics shortens shelf-life — when the named thing moves, the policy is stale.

Triggers & scope

In the last 6 months, did any of these change without a governance update?

Select all that changed with no governance review. Selecting any scores this section as a missed trigger.

Your policy was written for a deployment size / use case / jurisdiction — does that still match today?

Feedback & edge cases

If your AI governance were failing right now, which metric would tell you — and who watches it?

In the last quarter, how many "weird" / edge cases did your governance process actually surface?

How can a person affected by an AI decision surface a problem that changes the system?

Speed & escalation

Fastest your governance can make a binding change?

Is there one person with both the authority to pause a system and the technical depth to know when to?

Operational

How is a new AI tool request reviewed, and how long from ask to decision?

Do you have an inventory of AI-driven decisions (not just systems) — who made each call, when, why, and the reversal condition?

Has any governance control been re-examined on a schedule tied to outside events (not just when the system changed)?

Scored in your browser. Nothing is sent.

If scoring doesn't appear, your browser has JavaScript turned off — the button will instead open an email to me with your answers, and I'll reply with your read.

What this is, and isn't

  1. Instant, honest, and capped.

    A self-serve form can only check what you report, not the evidence behind it — so the read is honest about its own ceiling and never claims more certainty than your answers support.

  2. The paid Shelf-Life Assessment is the human-reviewed version.

    It gives every control a dated shelf-life and a re-examination calendar, takes seven to ten business days, and is reviewed by me — not generated. The delay there is the quality signal. See the Tier 2 engagement.

  3. No marketing list.

    Nothing here signs you up for anything. If you email your read, that email is retained until you ask to delete it. See the privacy notice .

Curious about the underlying framework? Read the methodology . It explains the eight structural causes behind documented governance failures, and why durability is a separate question from whether you pass a compliance check today.