AI Is Not the Strategy
How Leaders Turn Machine Intelligence into Business Value
The uncomfortable truth for every leadership team is this: artificial intelligence can make an organisation faster without making it any smarter. Across the boardroom, budgets are rising, copilots are being switched on, agents are being tested, and data platforms are being modernised. Employees are adopting tools faster than corporate policies can keep up. But beneath all that activity sits the question that matters: are we making better decisions? If not, AI may simply help the organisation repeat its mistakes faster, more efficiently and at greater scale. McKinsey’s August 2026 State of AI survey makes the gap clear. Four-fifths of respondents said AI had improved their individual productivity, and half said it had improved their decision-making. Yet only 37 per cent attributed at least some EBIT impact to AI use — essentially unchanged from the previous year. Just 6 per cent were McKinsey’s AI ‘high performers’. That is the gap executives must close. Productivity is not transformation. Adoption is not value and technology is not a strategy.
The real AI problem is not the technology.
For years, organisations have invested in technology to respond to disruption: ERP, cloud, data, analytics, automation and digital platforms. AI is now the latest addition. But technology rarely fixes a broken operating model. If decisions are slow, accountability is unclear, data is unreliable, and departments work in silos, AI won't fix those weaknesses. It may amplify them. A large language model cannot make an organisation intelligent when its data is fragmented, and its decision rights are uncertain. That is why one of the most dangerous questions a leadership team can ask is, ‘Where can we use AI?’ Start somewhere else: ‘Which business outcomes do we need to improve? Which decisions drive them? What stops those decisions being made well today?’ That moves the conversation from tools to value. It also exposes an uncomfortable reality: many of the barriers between AI investment and business performance have little to do with AI. They are poor information architecture, conflicting metrics, weak governance, technical debt, outdated processes, incentives that reward local rather than enterprise outcomes, and employees without the data, authority or confidence to challenge assumptions. AI does not solve those problems. It reveals them.
Four major consulting firms are telling leaders the same thing.
Strip away the jargon, and the message from McKinsey, BCG, Deloitte, and Accenture is remarkably consistent: the companies getting value from AI aren't simply deploying more technology. They are redesigning how work gets done.
McKinsey: Individual productivity is rising faster than enterprise value
According to McKinsey’s 2026 research, almost nine out of ten respondents used AI regularly in at least one business function. Forty-four per cent said their organisations were scaling AI across the enterprise. However, the financial impact remains limited. McKinsey found that high-performing organisations behave differently. Nearly three-quarters of its AI high performers said they had completely redesigned workflows around AI, compared with just a quarter of other respondents. High performers also combine efficiency goals with growth and innovation, rather than treating AI primarily as a cost-cutting tool. The implication is significant: the advantage lies not in applying AI to outdated processes, but in asking whether those processes should continue to exist.
BCG: AI is saving time — but businesses are wasting the opportunity
Another caution comes from BCG’s 2026 AI at Work research. Among frontline employees who use AI regularly, 42 per cent said they saved at least eight hours per week — the equivalent of a full working day. That appears transformative. Yet 66 per cent said they received little or no guidance on how to use the time AI saved, while more than half were not reinvesting that capacity in more strategic work. In other words, organisations may invest heavily in AI and save significant employee hours, yet still fail to create meaningful value if they do not redesign work to use those savings. BCG’s conclusion is especially direct: strategic clarity matters more than simply providing tools. Its 2026 survey found that clear strategic direction was associated with greater AI impact, even where access to AI tools was more limited. This prompts a more relevant question for leaders: instead of asking, ‘How many people use AI?’, ask, ‘Is AI helping people do valuable work differently?’ These are not equivalent measures.
Deloitte: Efficiency is winning. Reinvention is lagging.
Deloitte’s State of AI in the Enterprise 2026 reaches a similar conclusion. AI is delivering efficiency and productivity gains, but only 34 per cent of surveyed organisations said they were using it to transform their businesses deeply by creating new products and services, reinventing core processes or fundamentally changing their business models. The challenge grows as organisations move from question-answering AI to agentic AI—systems that can take action and carry out sequences of work with some autonomy. Deloitte’s research indicates that only one in five organisations has a mature governance model for autonomous AI agents. This extends beyond IT governance. It raises a leadership question: who decides what? When an AI system recommends a price change, refuses a transaction, prioritises a customer, changes inventory or produces a strategic recommendation, where does machine authority end and human accountability begin? These boundaries must be designed intentionally, not allowed to emerge by chance after deployment.
Accenture: AI is changing how knowledge moves through the business
Then there is the human dimension. Accenture’s 2026 Pulse of Change research found that AI adoption is nearly universal, while many organisations still see limited realised enterprise value. It also found that employees report tangible benefits from AI, including increased productivity and job satisfaction. This presents a more nuanced reality than the common claim that AI will eliminate jobs. AI is changing how knowledge moves within organisations. Previously, an employee facing an unfamiliar problem might have turned to a manager, colleague or subject-matter expert. Increasingly, they may start with a machine. This shift has major implications for mentoring, organisational learning, expertise and culture. If people increasingly turn to AI rather than colleagues, leaders must ask: what happens to apprenticeship? How do junior employees learn judgement? Where does tacit organisational knowledge come from? How do we learn to recognise when an apparently confident AI response is wrong? And perhaps most importantly, what capabilities become more valuable when information itself becomes cheap?
As machine capability rises, human judgement matters more.
This is where the technology conversation becomes human. AI is highly effective at processing information, identifying patterns, generating alternatives and accelerating work. But optimisation and judgement are not the same. AI strategist Sol Rashidi makes this distinction particularly well. Drawing on 13 years of experience and more than 200 enterprise AI deployments, Rashidi argues that AI will find the fastest path. Still, it lacks the relationship history, political context, ethical weight or lived experience needed to determine whether that path is right for a specific situation. Her warning is clear: the fastest solution is not always the right one. That distinction should resonate in every boardroom.
Suppose there is a restructuring. An AI system could calculate the organisational configuration that would produce the greatest theoretical efficiency gain. But should it determine what capabilities the organisation may need in five years? Does it understand the effect of removing a well-respected manager who informally keeps the team together? Can it evaluate the customer relationships held in an employee’s head rather than in the CRM system? Can it account for the reputational costs of a decision? The model can offer evidence. Leadership remains responsible for the final decision. In the wider context of Rashidi’s argument, organisations should explicitly decide what work should be human-led, what can be AI-led, what can be fully automated and where human–AI collaboration is appropriate. This is far more productive than simply instructing employees to ‘use AI more’.
A simple metaphor for every executive
Entrepreneur and investor Naval Ravikant recently extended Steve Jobs’ famous description of the computer as ‘a bicycle for the mind’, saying: ‘Maybe we have a motorcycle for the mind, to stretch the analogy, but you still need someone to ride it, to drive it, to direct it, to hit the accelerator, and to hit the brake.’ Ravikant isn't suggesting AI is unimportant; quite the opposite. A motorcycle greatly expands what a person can do, but it still requires someone to ride, steer, accelerate and brake. This model helps executives who don't consider themselves technologists. They do not need to be machine-learning engineers. They need to understand where AI should enhance human capability and where human oversight remains necessary. The executive skill therefore shifts from having all the answers to fostering an organisation that can ask better questions.
AI makes trusted data more important, not less.
The excitement around generative AI has sometimes made traditional data disciplines seem less important. In fact, the opposite is true. AI increases the value of reliable organisational information because it makes information radically easier to consume and act upon. However, if the underlying information is inconsistent, incomplete or poorly governed, the organisation can distribute those weaknesses at unprecedented speed. Data architecture is therefore not merely an IT concern. It determines whether people across finance, operations, marketing, supply chain and customer service make decisions from a common version of reality. Imagine an AI assistant preparing tomorrow morning’s executive briefing. It can summarise a thousand documents in seconds. Impressive. But what if finance defines “customer” differently from marketing? What if regional systems calculate margin differently? What if supply-chain information is three days late? What if no one knows which data source is authoritative? At that point, the model is not the issue; the challenge is organisational. Artificial intelligence cannot create a culture of facts from contradictory facts. Before organisations can become truly AI-enabled, they still need the less glamorous foundations: trustworthy data, sensible architecture, clear ownership, good governance and shared definitions of performance.
The C-suite must move from sponsoring technology to designing decisions.
For many executives, digital transformation once meant sponsoring a large technology programme and holding the CIO accountable for delivery. AI makes that model inadequate. It changes how decisions are researched, challenged and executed — and, inevitably, how the enterprise operates. AI is therefore a CEO, CFO, COO, CHRO and business-unit issue, not merely a CIO issue. The leadership team must be explicit about five things:
Outcomes: What measurable business outcomes should AI improve?
Decisions: What high-value decisions drive those outcomes?
Authority: Which decisions should humans make, which can be delegated, and where is collaboration appropriate?
Data: What data can those decisions rely on?
People: What should employees do with the capacity, insight and speed AI creates?
At this point, technology strategy becomes business transformation. Many organisations then discover their core challenges aren't about AI itself. They have problems with accountability, data, process or management.
AI has not created these problems. It has made them impossible to ignore.
Stop measuring AI adoption. Start measuring changed outcomes.
A dashboard showing that 70 per cent of employees have logged into an AI tool may look reassuring. It says almost nothing about transformation. Leaders need measures tied to business outcomes. Is the time needed to approve a customer proposal falling? Are forecasts improving? Are employees solving problems without escalating them? Are product development cycles shorter? Are managers spending less time gathering information and more time making decisions? Is customer churn down? Is working capital improving? Are employees using AI-created capacity to solve more valuable problems? These measures separate activity from impact. ROI should not be reduced to headcount savings. Removing labour from a process can deliver an immediate financial benefit. Redesigning the process can create an entirely new source of value. Leadership must move beyond automation towards augmentation, innovation and reinvention.
The human advantage is changing.
AI will certainly disrupt the work humans do today. The key leadership question is: what should humans do next? When machines can retrieve information almost instantly, knowing where to find it matters less; knowing whether it makes sense matters more. When machines can generate ten strategic options in seconds, options become less scarce and sound judgement becomes more valuable. When machines can cheaply produce competent presentations, reports, code and analysis, average output becomes abundant and discernment gains value. So do judgement, curiosity, empathy, accountability, imagination, courage and the ability to define the right problem. This is the irony of AI: as technology advances, distinctly human leadership capabilities become increasingly important. Not because humans must compete with machines at machine tasks, but because somebody must decide what is worth doing.
The next transformation is not digital. It is decisional.
For the past decade, organisations have talked about digital transformation. A more relevant concept now is decision transformation. After all, every business is a collection of decisions: who to target as customers, where to invest, what to build, what not to do, whom to hire, which risks to take, whom to trust and when to reject conventional wisdom. AI can provide better information for those decisions. It can expand the range of possibilities, simulate scenarios, uncover patterns no executive could sift through manually, challenge assumptions and accelerate execution. However, technology does not absolve leaders of responsibility; it increases it. Companies that sustain a durable advantage from AI are unlikely to be those with the most licences or the most impressive demos. They will have trustworthy information, sound architecture, redesigned processes, clear decision rights and employees who know when to trust the machine and when to challenge it.
The one question every executive team must ask
The next time an AI initiative comes before the board, don't start with the model, the vendor, or the number of agents you can deploy. Begin with one question: Which decision or business outcome will this initiative materially improve? Then ask: What data will that decision rely on? Is it trustworthy? Who has the authority to act? Where can AI help? Where must human judgement remain? What will be different as a result? How will we know we have created value? These questions may be less exciting than a demonstration of the latest AI agent, but they are far more likely to determine whether the investment pays off. AI will be everywhere. Simply using it will not set a business apart. The winners will be the organisations that combine machine intelligence with human judgement, trusted information and decisive leadership. Artificial intelligence can turbocharge your organisation. But someone must still choose the destination — and lead the way.
Sources: McKinsey & Company, The state of AI in 2026: On the road to ROI, 25 August 2026; Boston Consulting Group, AI at Work: Strategy Matters More Than Tools, 3 June 2026; Deloitte, The State of AI in the Enterprise: The Untapped Edge, 2026; Accenture, Pulse of Change, 26 July 2026; Sol Rashidi, AI Efficiency vs Human Judgement in Decision Making, 2026; Naval Ravikant, A Motorcycle for the Mind, 2026.
