Overview

The Board AI Readiness Check is an 18-question anonymous diagnostic. Each item maps to one of six dimensions of the STAR framework, extended for AI-era board governance. Every item has four forced-choice options scored 0 to 3, with no midpoint. Total scores range 0 to 54 and map to one of four tiers, subject to capability gates that prevent a board from claiming a tier it lacks the machinery to defend. Two derailment flags surface risky patterns independent of tier. The Check captures no name, email, or organisation at any point.

The six dimensions

S. Shareholder value thesis. Whether AI is connected to how the company creates value, and whether the board owns the thesis rather than borrows it.

T. Threat parity. Whether the board is matching the speed and capability of the fastest movers, and has stress-tested the business against AI-enabled disruption.

A. Ability. Whether the company can execute what it is describing: real workflows, real talent, real data foundation. This dimension measures the company’s capacity, not the board’s.

R. Risk budget. Whether AI risk is defined in writing, owned by name, and monitored after launch, with a documented trail if something fails.

O. Ownership. Where AI oversight is written down: in a committee charter or board mandate, in a board-approved policy, and in a defined escalation path measured in days.

Au. Augmentation. The board’s own AI capability: distributed director fluency, information reaching the board when facts change rather than when the calendar allows, and directors using AI to test what they are told.

The eighteen items

S. Shareholder value thesis

S1. Can the board state, in one sentence, how AI is expected to change this company’s value creation over the next three years?

  • No, it has not been framed that way. (0)
  • Roughly, but the answer differs depending on who you ask. (1)
  • Yes, leadership has a thesis the board has heard. (2)
  • Yes, and the board can also say what would change it. (3)

S2. Who owns that thesis?

  • Nobody, it does not exist in a form anyone owns. (0)
  • Management holds a view the board has not examined. (1)
  • Management owns it and the board has reviewed it. (2)
  • The board has tested and shaped it, and revisits it when the facts move. (3)

S3. How does the thesis show up in capital allocation?

  • It does not. (0)
  • AI spending happens, but outside any AI-specific investment case. (1)
  • Major AI investments carry business cases the board reviews. (2)
  • AI investment is sized against the thesis, with expected return, milestones, and stopping rules. (3)

T. Threat parity

T1. Does the board know, in specific and current terms, what the company’s most aggressive competitors are doing with AI?

  • No. (0)
  • Vaguely, from press and anecdote. (1)
  • Yes, at a high level, as presented by management. (2)
  • Yes, in specific and current detail, including from sources outside management. (3)

T2. Compared with the fastest movers in the sector, where is this company on real AI deployment?

  • Behind, and not closing the gap. (0)
  • Behind, and aware of it. (1)
  • Roughly at parity. (2)
  • Ahead, and treating it as a defensible advantage. (3)

T3. Has the board stress-tested the business against AI-enabled disruption, whether a lower-cost AI-native competitor or an AI-enabled attacker?

  • We have not considered it. (0)
  • We have named the risk but not planned against it. (1)
  • We have a general response in mind. (2)
  • We have stress-tested it and have moves ready. (3)

A. Ability

A1. Does the company have the talent and tools to execute its AI ambitions, as opposed to slideware?

  • No, there is a clear capability gap. (0)
  • Partially, with notable holes. (1)
  • Mostly yes. (2)
  • Yes, and it is a deliberate strength. (3)

A2. Is AI in real use in real workflows?

  • Licences and pilots, little real use. (0)
  • Some pilots, nothing at scale. (1)
  • In production in several workflows. (2)
  • Embedded in core workflows, with measured impact. (3)

A3. Does the company have the data foundation its AI ambitions require?

  • No, data is fragmented and quality is unknown. (0)
  • Partially, with known gaps. (1)
  • Mostly, in the areas that matter. (2)
  • Yes, and data quality is measured and owned. (3)

R. Risk budget

R1. Has the board defined how much AI-related risk this company is willing to accept, in plain terms?

  • No risk appetite has been set. (0)
  • Informally, nothing written. (1)
  • Yes, at a high level. (2)
  • Yes, specific, written, and reviewed on a schedule. (3)

R2. Does the board know which AI uses inside the company are high-risk, and who owns each one?

  • No, we do not have that picture. (0)
  • Partially. (1)
  • Mostly, with named owners. (2)
  • Fully inventoried, with named owners and monitoring after launch. (3)

R3. If an AI system caused a public failure tomorrow, could the board show what it did beforehand?

  • No, we would have hope. (0)
  • We would be partly covered. (1)
  • Largely, yes. (2)
  • Yes, with a documented trail that would stand up to a court or a regulator. (3)

O. Ownership

O1. Where is AI oversight assigned on this board?

  • Nowhere, it has not been assigned. (0)
  • Informally, it comes up wherever it lands. (1)
  • Assigned in practice to a committee or the full board. (2)
  • Written into a committee charter or board mandate, with no AI duty left unowned. (3)

O2. Does the company have an AI policy, and has the board seen it?

  • No policy. (0)
  • A policy exists somewhere, the board has not reviewed it. (1)
  • Yes, and the board has reviewed it. (2)
  • Yes, board-approved, reviewed on a schedule, and tested against actual practice. (3)

O3. If something went wrong with an AI system, how fast would the board hear?

  • We would learn publicly, or at the next scheduled meeting. (0)
  • Through the normal quarterly reporting cycle. (1)
  • There is an escalation path, though the timing is undefined. (2)
  • A defined escalation path measured in days, and it has been tested. (3)

Au. Augmentation

Au1. How would you rate the board’s own AI fluency, not management’s?

  • Low, most directors could not lead the discussion. (0)
  • Mixed, one or two carry it. (1)
  • Solid across most of the board. (2)
  • High, distributed rather than concentrated in one director, and refreshed on a schedule. (3)

Au2. Between meetings, how does the board learn that something material has changed?

  • It does not, we find out at the next meeting. (0)
  • Ad hoc, if management chooses to raise it. (1)
  • There is a standing update between meetings. (2)
  • Information reaches directors when the facts change, including from outside management. (3)

Au3. Do directors use AI themselves to prepare and to test what they are told?

  • No. (0)
  • One or two do, informally and privately. (1)
  • Several do, and it is accepted practice. (2)
  • Yes, working from a common factual base with each director applying their own lens. (3)

Scoring and tier thresholds

Each dimension sums three items and caps at 9. The total is the sum of all six dimensions and caps at 54. Base tier is assigned from total:

  • 0 to 17: Flying Blind.
  • 18 to 36: Briefed.
  • 37 to 48: Equipped.
  • 49 to 54: Augmented.

Capability gates

A board cannot claim a tier it lacks the machinery for. Two dimensions act as gates: Ownership (O) and Augmentation (Au). Both are dimensions because a board that scores well on strategy but has not written oversight down, and cannot independently test what it hears, is still doing attendance rather than oversight.

Equipped tier requires O ≥ 5 and Au ≥ 5. If either gate fails, the tier drops to Briefed and the gate message on the result explains which capability is missing.

Augmented tier requires O ≥ 7 and Au ≥ 7. If either gate fails at this threshold, the tier drops to Equipped and the gate message names the missing capability.

When both gates fail at the same threshold, Ownership takes precedence in the message shown, because architecture comes before capability.

Derailment flags

Two independent flags surface risky patterns that the tier score alone can miss. They can co-occur across different dimensions.

Moving without guardrails. Triggered when A ≥ 7 and R ≤ 4. The company has capability but the board has not sized the risk. This combination is not a stage on the way to good governance, it is a way off the path.

Governed but losing. Triggered when R ≥ 7 and T ≤ 4. The board has built a brake rather than power steering. Control without speed is its own exposure, because the competitive threat does not pause while the board gets comfortable.

Conditional callouts

Two additional questions are surfaced when a board’s scores make them newly relevant.

Privilege and discovery of AI-assisted board materials. Shown when Au ≥ 7. Boards using AI well are the ones for whom the 2026 case law on AI-assisted work product matters first.

Incentives tied to AI value creation. Shown when S ≥ 7. A clear value thesis makes it possible to ask whether management is paid for launching AI or for the value it produces after the cost of controls and the risk it carries.

Softest dimensions

After the tier and grid, the results screen names the two lowest-scoring dimensions as the natural starting point for the next board conversation. Ties break in the order S, T, A, R, O, Au. That order is deliberate: strategy first, then threat, then capability, then risk, and finally the two capability gates that already carry their own emphasis.

What is captured, and what is not

Every completed Check is captured to a database at the network edge: the tier, the score for each dimension, the answers array of 18 integers, the two flags, the gate outcome, the source page, the country from the edge network, and the instrument version. The database retains no name, email, or organisation, because the Check never asks for any of them. The results screen carries no capture form of any kind.

This is a deliberate design choice, not an omission. The Check is a diagnostic, not a lead form. Boards should be able to see where they stand without paying for it in contact detail.

Changelog

2.0, July 2026. Expanded from 12 items and four dimensions to 18 items and six. Added Ownership, covering where AI oversight is written down, whether an AI policy exists and has been tested, and how fast the board hears. Added Augmentation, covering the board’s own fluency, information flow between meetings, and director use of AI. Removed a duplicated competitive-threat item. Moved board fluency out of Ability, which now measures the company’s capacity to execute rather than the board’s capacity to know. Added capability gates so a board cannot reach a tier it lacks the architecture for, and two derailment flags. Version 1.0 submissions are retained and excluded from version 2.0 aggregates.

1.0, May 2026. Initial release. 12 items across four STAR dimensions (S, T, A, R), four tiers based on total score with no gates.

Take the Check

Start the Board AI Readiness Check. Five minutes. No email required.