That page is the real change. An Intelligent Enterprise is not one that has bought more AI. It is one whose board can see how its decisions are made, by people and by software, and can widen or narrow that delegation with confidence.
Where most boards are today
The distance to that meeting is still considerable. The 2026 Gartner CIO and Technology Executive Survey, reported in Gartner’s 15 April 2026 article on agentic AI, found that AI agents were already deployed in 17% of organisations; more than 60% expected deployment within the following two years. McKinsey’s survey published on 25 August 2026 found that 37% of respondents reported enterprise-level EBIT impact from AI.
Our view is that governance will determine how much of this investment becomes dependable business value. A board needs to know which decisions it is prepared to delegate, on what evidence, and how it will know when something has gone wrong.
We have spent our careers on the systems enterprises run on: a finance platform serving 153 countries, ten factories brought onto one ERP, programmes where financial accuracy carried consequences for the whole operation. The lesson carries straight into AI. A system earns authority by being accountable, not by being clever.
Three shifts that define the governed enterprise
From reports to a decision register. Boards today see outcomes. In the Intelligent Enterprise they will also see decisions: which were taken by people, which by software within delegated limits, and the evidence behind each. The register becomes the board's instrument for AI in the way the risk register became its instrument for risk. The US National Institute of Standards and Technology already treats AI governance as a thread through the whole lifecycle rather than a gate at the end. Boards will apply the same idea to the business decisions AI touches.
From approving projects to granting authority. Investment committees currently approve AI projects. The more consequential question is what authority each capability holds. Software that drafts a supplier recovery plan is a different proposition from software that places the order. We expect boards to manage AI delegation much as they manage financial delegation: tiers, limits, named owners and a clear way to withdraw authority as soon as the evidence weakens. Autonomy is earned, not assumed.
From annual assurance to continuous evidence. Internal audit will sample AI decisions as it samples transactions, reconstructing a handful each quarter from the information used, through the recommendation and the human response, to the outcome. Directors will not need to understand models. They will need to see that the organisation can explain any consequential decision on request.
What this makes possible
Get the governance right and the enterprise moves faster, not slower. Management can let software settle routine exceptions overnight, because the board agreed the limits in advance. Capital becomes more reversible: a capability that stops earning its keep is retired on evidence rather than defended out of pride. Changing an AI supplier becomes an operating decision rather than a crisis, because the business rules and decision records sit outside any one product.
As a result, the board's attention moves to where it belongs: the decisions it has chosen to keep.
The same principle, different pressures
The principle travels; the pressure differs by market. On 23 April 2026, the UAE announced a framework targeting agentic AI across 50% of federal government sectors, services and operations within two years. That ambition puts delegation on the operating agenda. In Europe, the AI Act establishes duties for high-risk systems, including records and human oversight; boards need to prepare for the applicable requirements and implementation dates. A decision register can support that work. In India, the Digital Personal Data Protection Act sets the framework for lawful processing of digital personal data. The board needs a clear account of which information its agents use and why.
Put the decision register on this quarter’s agenda
Ask management for one page: the ten most consequential recurring decisions in the business, who takes each today, and which could be prepared or taken by software. Then agree the evidence the board would need before delegating the first of them.
That page is the beginning of your decision register, and the first agenda item of the 2030 board. If you would like to work through it with us, a managing partner will take your executive team through it in a working session.