— 01Strategy
Can your CEO state — in one sentence, without the deck — what AI changes about how this company makes money?
A clear one-sentence answer suggests strategy. A list of pilots suggests theatre.
— Diagnostic · v1
The AI Readiness Audit.
Ten questions, fifteen minutes. Locates where your firm sits on the path from AI-as-feature to AI-as-foundation. Score yourself, forward it to a colleague, take it into a board conversation.
This is not a marketing form. It does not collect your answers. Print it, fill it in, sit with the result. If you'd like a working session on what to do next, the closer at the bottom is the door.
— Pairs with the essay AI is not a feature →
— 02Operating model
Has anyone outside the executive team been given permission to say no to an AI initiative on technical grounds?
Authority to decline is a leading indicator of governance. Absence of that authority is a leading indicator of theatre.
— 03Data foundation
If your CFO asks for revenue by segment, does she get the same number twice from the same system?
If yes, you have a semantic layer. If no, your AI work will surface that gap before it surfaces value.
— 04Platform
Can a team ship a new AI feature to production this week if the business decided this morning?
The answer measures your platform's posture far better than a vendor's slide measures it.
— 05Evaluation
For your most consequential AI feature, is there a published eval set the team would defend in a regulator's room?
If the eval is private, the feature is exploratory. If the eval is public to the team, the feature is production.
— 06Production
When a production AI behaviour drifts, how long before someone is paid to notice?
Detection latency is the most useful production metric for AI systems. Most firms don't measure it.
— 07Talent
How many people in your firm — outside the data and AI teams — are using modern AI tools in their daily work?
Diffusion through the firm matters more than depth in a single team. The leap is everyone's, or it's nobody's.
— 08Governance
If a regulator asked tomorrow how an AI decision was made for a specific customer, could you answer within a week?
Explainability isn't a slide. It's an architecture choice you make eighteen months before the regulator calls.
— 09Compounding
Did anything your AI team built in the last quarter make the next thing easier to build?
Capability that compounds is the signal of a firm shifting into the next level. Capability that's one-off is a sign of pilots.
— 10Conviction
If the AI work disappeared tomorrow, would your strategy still make sense — or was AI the centre of it?
A firm where AI is the substrate sees the strategy collapse without it. A firm where AI is a feature sees the strategy survive. Either answer is fine; the wrong answer is not knowing.
Reading the total.
Add your scores. The maximum is 50. The number on its own means less than the shape of where you scored highest and lowest — that's where the next twelve months of work probably lives.
- 10 — 20 Pre-shift. AI is a feature, not a foundation. The next twelve months should be about thesis and foundations — a clear one-sentence strategy, a semantic layer, an eval framework. Build less. Decide more.
- 21 — 30 Mid-shift. You have pilots, a partial platform, and the beginnings of governance. The risk is plateau. The next move is to compound — pick one consequential surface and make it Level-03 architected, not Level-02 augmented.
- 31 — 40 Advanced. You are nearly AI-architected. The remaining work is usually about evaluation discipline, governance, and cross-firm diffusion. The leap is in sight; the last mile is mostly cultural.
- 41 — 50 Already AI-native — you should probably be writing this audit, not taking it. We'd be interested to compare notes.
— No score is a verdict. The diagnostic is a mirror, not a grade.