APRIL 28, 2026
AI is no longer a future conversation. It is already being implemented, tested, and in many cases operationalised across industries. From financial services automating compliance checks, to retailers using AI agents to manage inventory in real time, to professional services firms rebuilding how they deliver client work entirely.
But while the narrative continues to accelerate, the reality on the ground is more nuanced. Some businesses are moving quickly and effectively, compounding advantages with every quarter. Others are investing heavily, running pilots, announcing AI strategies, and seeing limited return. For founders and business leaders exploring AI right now, this raises a more important question than whether to adopt it: why are some organisations seeing real impact, while others are not?
The answer, more often than not, has nothing to do with the technology.
On paper, adoption looks strong. According to McKinsey, 88% of organisations now use AI in at least one business function, up from 72% just two years ago. Generative AI alone has gone from 33% adoption in 2024 to 72% in 2025. By almost every measure, AI has crossed from early adoption into mainstream business use.
But only around one-third of those organisations are scaling it meaningfully across the enterprise. The rest are caught in what many are now calling “pilot purgatory”, running proof of concepts that demonstrate promise but never translate into operational reality.
The gap is not access to technology. It is execution. And in many cases, it comes down to a fundamental misunderstanding: are we implementing AI, or simply improving automation? These are not the same thing. Automation replaces a manual step with a faster one. AI done properly changes the logic of how work gets done altogether. Confusing the two is one of the most common and costly mistakes businesses make at the start of an AI programme, and it tends to compound over time.
As businesses move quickly to adopt AI, the focus often sits on tools rather than foundations. There is a natural pull toward the visible: the product demos, the vendor promises, the competitor announcements. But AI does not operate in isolation. It relies entirely on how the organisation is structured, how data flows, and how decisions are made. Drop even the most advanced model into a poorly structured environment and it will produce unreliable outputs, create confusion about ownership, and ultimately erode confidence in AI across the business.
This is why a small number of organisations are pulling ahead. Not because they have access to better technology, but because they have built the right environment for it to work.
Across those organisations, three patterns consistently emerge.
The first is a clear and governed data foundation. AI is only as good as the data it operates on. Organisations seeing consistent results have defined data ownership, structured access, and quality controls before any model goes near production. This is not glamorous work, but it is foundational.
The second is genuine workflow redesign. Not layering AI onto existing processes, but interrogating how work actually moves through the business and rebuilding around what AI makes possible. This is the single strongest predictor of financial impact. McKinsey’s data shows that high performers are nearly three times more likely to have fundamentally redesigned their workflows, and that this redesign, more than any other factor, determines whether AI delivers measurable business value.
The third is a defined operating model with clear ownership and human oversight built in. AI does not run itself. Someone needs to own the outputs, monitor performance, manage risk, and make decisions about when human judgement overrides the model. Organisations that establish this structure early scale with confidence. Those that do not tend to discover the gaps at the worst possible moment.
Most organisations do not start here. The pressure to show progress, to leadership, to boards, to the market, pushes businesses into implementation before the foundations are ready. And that creates a set of familiar, compounding problems.
Automation gets labelled as AI, creating an inflated sense of progress. Data remains fragmented across systems that were never designed to talk to each other. Governance is treated as something to address later, once the technology is working. Infrastructure and ongoing costs are underestimated, particularly as usage scales. And when results disappoint, the conclusion is often that the technology failed, when in most cases the environment around it was never set up for success.
It is worth noting that 51% of organisations have already experienced negative consequences from AI use. That figure rarely makes it into headline adoption statistics, but for founders making implementation decisions now, it is an important data point. Moving fast without the right foundations is not a competitive advantage. It is a liability that tends to surface at scale.
The nature of AI in business is shifting. We are moving from AI as a tool you interact with, to AI as something that actively operates within your business, taking actions, making decisions, executing workflows autonomously.
Around 62% of organisations are now experimenting with AI agents, and nearly a quarter are scaling them in at least one function. These are systems capable of multi-step, autonomous execution. Not answering a question, but completing a task end to end. In practical terms, this means AI that can manage a customer query from first contact to resolution, process and route a document through an approval workflow, or monitor a system and take corrective action without human intervention at each step.
This represents a meaningful shift in both opportunity and risk. Governance is already struggling to keep pace with capability. Security and risk concerns are now the top barrier to scaling agentic AI, ahead of regulatory uncertainty and technical limitations. Only about 30% of organisations have reached meaningful maturity in AI oversight. The capability is running ahead of the controls, and for founders making early implementation decisions, that is a risk to account for from the very beginning rather than something to address once problems emerge.
For most organisations, the right starting point is an honest assessment of where they actually are. Not where they hope to be, and not benchmarked against the most advanced organisations in their sector. That means auditing data quality and ownership before evaluating any tools, mapping existing workflows to identify where AI genuinely adds value rather than where it simply sounds impressive, and defining what success looks like in specific, measurable terms before a single pound is committed to implementation.
From there, three things consistently separate the organisations that scale from those that stall.
Governance before deployment. Define who owns AI outputs, how accuracy is monitored, what the escalation path looks like when something goes wrong, and when human oversight is required. This structure does not need to be complex at the outset, but it needs to exist before rollout, not be retrofitted after an incident. Organisations that treat governance as an afterthought tend to find that trust in AI, internally and externally, is much harder to rebuild than it is to establish from the start.
Workflow redesign over tool adoption. The most important question at the start of any AI programme is not “which tool should we buy?” It is “how does this change the way work actually gets done?” These are fundamentally different questions, and they lead to fundamentally different outcomes. Buying a tool and expecting it to transform a workflow is like buying gym equipment and expecting to get fit without changing your routine. The organisations seeing the strongest returns are the ones that treat AI as a reason to interrogate their processes, not just accelerate them.
Leadership that does more than sign off. In high-performing organisations, senior leaders do not simply approve AI budgets and wait for results. They actively champion adoption, role-model usage across the business, and tie AI initiatives to specific outcomes with clear KPIs. They create the conditions for change, not just the permission for it. McKinsey identifies senior leadership engagement as one of the strongest individual predictors of AI success, and it shows up consistently across every sector and company size.
Getting these three things right requires a combination of the right people, the right structure, and clear delivery accountability, particularly at the point when AI moves from exploration into execution. That transition is where many organisations stall, not because of a lack of ambition, but because the internal capability to deliver at that level is not always there. Knowing when to bring in specialist support, and choosing partners who understand both the technical and organisational dimensions of AI delivery, is itself a strategic decision.
The organisations that will benefit most from AI are not necessarily those investing the most. They are the ones willing to slow down at the start, to build the foundation properly, so that when they scale, it holds. That discipline is less common than it should be, but it is the single clearest differentiator between organisations that realise lasting value and those that accumulate technical debt and diminishing returns.
For founders considering AI right now, the question is no longer whether to adopt it. It is how to do it in a way that creates structural advantage rather than just visible activity. That means being honest about current capability gaps, making deliberate decisions about where specialist expertise is needed, and treating AI as a long-term operating model decision rather than a technology purchase.
The businesses getting this right are not doing it alone. They are partnering with people who have delivered this before, who understand the foundations, the risks, and the path from pilot to scale. That combination of specialist talent and structured delivery is what turns AI ambition into operational reality.
