Your Strategy Is AI-Ready. Your Organisation Isn't.

Most executive teams can describe their AI strategy on a single slide. The ambition is clear, the use cases are shortlisted, and the technology partners are chosen. Then the work begins, and progress slows in places the strategy never mentioned.

This is the AI readiness gap. It sits between what a strategy says the organisation will do and what the organisation's structure, workflows, leadership habits and skills can actually support. The gap is rarely about technology. It is about whether the organisation around the technology has been prepared to change, and this article sets out where that preparation is usually missing and how an executive team can check for it.

What does it mean when a strategy is AI-ready but the organisation is not?

It means the plan is sound on paper, but the organisation cannot yet carry it out. A strategy defines where AI will create value. Organisational readiness decides whether the people, structures and processes are in a position to deliver that value. When the two are out of step, pilots succeed in isolation and then stall when they meet the real business.

A Harvard Business Review article by Cyril Bouquet, Christopher Wright and Julian Nolan gives a clear example. In 2018, General Motors used generative design to produce a seat bracket that was 40% lighter and 20% stronger than the original. It was never manufactured, because the company's production system was built around stamped steel and could not handle the new geometry. Updating that system would have taken years.

The technology worked. The organisation could not absorb it. The authors argue that AI ambition has to fit an organisation's operating reality, and that many bold pilots fail because it does not. That is the pattern this gap describes, and it applies well beyond manufacturing.

Why do executive teams overlook the gap?

Strategy and readiness are measured differently, and only one of them is reviewed regularly at the top. Strategy has milestones, budgets and named vendors. Readiness has no natural owner, so it rarely gets the same scrutiny.

Three habits tend to keep the gap out of view.

• Distance from the work. Senior leaders usually see AI in demonstrations and summary reports. They see less of the daily friction, such as unclear approvals, unowned handoffs and teams unsure whether they are allowed to change a process.
• Readiness treated as a technical question. Data quality, platforms and security are checked carefully. The human side of adoption is assumed to follow.
• Different clocks. Strategy runs on annual cycles. Behaviour, trust and skills change over much longer periods, so a plan can look complete long before the organisation is ready for it.

The numbers support this. McKinsey's 2026 state of AI survey, based on 1,719 participants in 97 countries, found that 44% of respondents say AI is scaling across their enterprise. Yet only 37% attribute any EBIT impact to AI, and just 6% qualify as high performers, meaning they attribute at least 5% of EBIT to AI and describe the impact as significant. That share was unchanged from the previous year. Activity is common. Measurable return is far less so.

None of these habits reflects poor intent. Each is a reasonable response to how large organisations are set up, which is why the gap tends to persist even where leaders are well informed about AI itself.

Does the operating model support how AI changes work?

In many organisations it does not. Most operating models were designed around stable roles, fixed approval chains and clear functional boundaries. AI changes what a role contains, how quickly a decision can be made and who needs to be involved. If structure and decision rights stay as they were, AI gets added to old ways of working instead of changing them.

A few questions show whether the operating model is keeping pace.

• Who decides when an AI output conflicts with an experienced person's judgement?
• Where does accountability sit when AI-assisted work crosses several functions?
• Can the teams using AI change the process around it, or can they only suggest changes?
• Do role descriptions and performance measures still describe the work people now do?

When these questions have no clear answer, people fall back on caution. They check everything twice, escalate decisions that could be made locally, or use AI quietly and avoid mentioning it. Each of these protects the individual and slows the organisation.

Reviewing structure, decision-making and workforce design in light of AI is the work of organisational design. It is usually less visible than a technology rollout, but it determines whether the rollout can be used.

Have workflows been redesigned, or has a tool simply been added?

In most organisations a tool has been added. Workflow redesign means changing the sequence of work, the handoffs and the roles so that AI sits inside the process, not beside it. Adding a tool to an unchanged workflow tends to produce small, local time savings that never appear in business results.

The McKinsey survey points in the same direction. Nearly three-quarters of AI high performers report fundamentally redesigning workflows because of their AI use, compared with about a quarter of other respondents. The survey describes association, not proof of cause, but the difference between the two groups is large enough to be worth examining.

A simple way to tell the two situations apart is to look at what changed after the tool arrived.

• If the steps, approvals and handoffs are the same as before, a tool was added.
• If the steps were reordered, some were removed, and the roles around them were adjusted, the workflow was redesigned.
• If nobody can say which of these happened, workflow ownership or oversight may be unclear.

Workflow redesign takes longer than a tool deployment and needs input from the people who run the process every day. That input is often the part skipped when a strategy is handed down.

Do leaders behave in ways that make AI adoption workable?

Adoption follows what leaders do, not what they announce. If senior leaders do not use AI in their own work, ask for AI-supported analysis, or discuss where it went wrong, employees reasonably conclude that it is optional or risky. Leadership behaviour sets a kind of permission that no policy document can.

Several behaviours matter in practice.

• Visible use. Leaders who show how they use AI, including what they check and what they reject, make it easier for others to start.
• Tolerance for error. Teams experiment more when early mistakes are treated as information and not as failures.
• Honesty about roles. Silence about how AI will affect jobs does not prevent speculation. It leaves people to fill the gap themselves, usually with the more worrying version.
• Middle manager support. Managers translate strategy into daily priorities. If they are unsure what is expected, their teams will be unsure too.

These behaviours are often missing from an AI roadmap, which is partly why it is missed. It sits closer to leadership than to technology planning, because the question is how leaders lead through a change they are also still learning about.

Do people have the skills and confidence to work with AI?

Usually some do and some do not, and the distribution is uneven. Capability here means more than knowing how to operate a tool. It includes judging the quality of AI output, knowing when not to use it, explaining AI-supported decisions to colleagues and adapting as the tools change.

One-off tool training covers only the first of these. The others are closer to professional judgement, communication and adaptability, and they develop through practice on real work. Capability development that treats working with AI alongside those human skills is one way to approach this.

Uneven skills also create a quieter problem. A small group of confident users pulls ahead, while others avoid AI because they are unsure how to start or worry about being seen as unprepared. Over time the organisation has two working styles that do not fit together well, and handoffs between them become another point of friction.

Does organisation size change the picture?

Size changes the form the gap takes, not whether it exists. Larger organisations usually have more resources to scale AI, while smaller ones have fewer layers and can often decide faster. Each faces a different version of the same readiness question.

The McKinsey survey reports that 54% of organisations with at least $1 billion in annual revenue say AI is scaling across the enterprise, compared with about one-third of smaller organisations. Scale of deployment, however, says little about whether the organisation has absorbed the change. In a large company the gap often appears as coordination trouble between functions, where one team's redesigned process meets another team's unchanged one. In a smaller company it more often appears as dependence on a few capable individuals, which leaves AI readiness fragile when those people move on.

How can an executive team test its own AI readiness?

By asking plain questions about the organisation, then checking the answers with people who do the work and not only with the people who own the strategy. The aim is to find where the leadership picture and the day-to-day reality differ.

These questions are a reasonable starting set for an AI readiness review.

1. Which two or three workflows carry the most value in the strategy, and have they been redesigned?
2. Who owns each of those workflows from start to finish?
3. Who has authority to change how work is done when AI makes a better route possible?
4. How do senior leaders themselves use AI, and do their teams know it?
5. What have employees been told about how AI will affect their roles?
6. Which skills are missing, and are they missing across the organisation or only in some teams?
7. What happens when an AI-supported decision turns out to be wrong?

Run the same questions with senior leaders and, separately, with team managers and frontline staff, then compare the answers. Where they match, the organisation probably has shared understanding. Where they differ, the difference is itself useful, because it shows where the strategy and the lived experience of the organisation have separated.

What should be fixed first?

Start where the strategy meets daily work most directly, which is usually decision rights and one or two core workflows. Fixing these gives later investment in skills and culture something concrete to attach to. Working on culture in general terms, with no connection to a specific process, tends to stay abstract.

A workable sequence looks like this.

1. Choose a small number of workflows. Pick the ones the strategy depends on most, and leave the rest for later.
2. Clarify decision rights. Write down who decides what, including when AI output and human judgement disagree.
3. Align the leadership team. Agree what leaders will do visibly, and what they will say about roles and change.
4. Build capability around those workflows. Train people on the work they actually do, not on tools in the abstract.
5. Review after a defined period. Compare what teams report with what leaders expected, and adjust.

AI readiness is not a one-time state that an organisation reaches and then keeps. Tools change, roles shift and new questions about accountability appear. A strategy that is AI-ready gives an organisation a direction. Organisational readiness decides how far along that direction it can actually travel, and treating the two as separate pieces of work is what allows the gap to close.