For boards and audit committees
AI readiness questions every board should ask
From the outside, two companies can show the same revenue, margins, systems and AI plans — and still face very different futures. The book argues the difference is operating clarity: whether a company can explain how it operates, who owns its critical data and decisions, and whether its AI connects to real outcomes. It names that as something boards should measure, and its seven-question scorecard gives directors ready questions for management.
Kindle edition · 99 pages · a 90-minute readRead a free sample on Amazon
From the book
What the book shows
- 01
Boards are told what to measure
The book calls the capability AI readiness: the structure, shared meaning, ownership, governance and decision discipline needed to scale AI. A company that lacks it can still look modern — tools, pilots, demos — while depending on a few individuals underneath.
“All of this points to a capability worth naming, because it is what investors, boards, and CEOs should actually be measuring.”
Chapter 17Scaling AI on a Foundation of Clarity - 02
The board’s instinct can be the wrong one
In the book’s (anonymized) merger case study, integration had ground to a halt eighteen months in. The board’s instinct was to allocate another $12 million to a total system consolidation. What unlocked the stalled savings was a structural reset of definitions and named owners — not more integration spend.
“It made incremental progress on paper and failed to resolve the gridlock, because the barrier was never technical.”
Chapter 20Case Study: When Two Strong Companies Could Not Become One - 03
An error in a model is not an error in a report
As AI moves from reports into decisions, mistakes stop being local. Hence the book’s insistence on a named owner for every critical domain, and on governance judged by the ambiguity it removes rather than the forums it runs.
“An error in a report is inconvenient. An error embedded in a model is systemic.”
Chapter 11Ownership Is a Leadership Role
For audit and risk committees
The book is skeptical of governance born from failure — a breach, a wrong report, an audit finding — that responds by adding controls and forums. Data quality, it argues, must be owned where the data is created; as decisions become automated, governance should pin down who is accountable instead of stacking up reviews.
“Sustainable data quality cannot be audited into a system after the fact.”
Questions to ask
Questions a board can put to management, built on the book’s seven tests:
- 1Can management show, on one page, how the business actually operates?
- 2Do core terms such as customer, revenue and margin mean the same thing across the group?
- 3Does every critical data domain have a named business owner with real decision authority?
- 4How much executive meeting time goes into reconciling numbers rather than deciding?
- 5Which AI initiatives trace to a business decision and an owner — and which are experiments?
- 6Is a large system program presented as a precondition for AI? On what evidence?
- 7Have the executives scored the seven questions independently — and how far apart are their answers?
Where to start in the book
Chapter 17
Scaling AI on a Foundation of Clarity — what “AI-ready” means, and why boards should measure it.
Chapter 20
Case Study: When Two Strong Companies Could Not Become One — the merger where more system spending did not help.
Appendix
The AI-Readiness Scorecard — the scored version of the seven questions.
Kindle edition · 99 pages · a 90-minute readRead a free sample on Amazon
Also see
- Investors & M&AAI due diligence: questions about a target’s operating model
- CEOsAI readiness for CEOs: what can’t be delegated
- Risk & complianceAI governance without bureaucracy
All roles: who the book is for →
About this page: our reading of Your data isn’t broken. Your operating model is. by Adam Ayres and Behrang Karimi, for boards and audit committees. The book itself is written for the leadership teams of complex enterprises. Quotes are word for word from the book; its case studies describe anonymized companies, and their results are not benchmarks.