By Rebecca Pluthero
AI delivers lasting value when businesses correctly identify the problems, assign ownership early, and govern for real-world scale.
AI adoption is accelerating, but speed alone does not equal strategic progress. Planning, legal foresight, and synergy between stakeholders are crucial to develop an adoption path capable of agility and scale. In all of this, AI governance is an enabler for real transformation and growth.
Start with alignment, not tools
I am always wary of the term “AI strategy”. AI can be a powerful part of an IT or business transformation strategy, but when AI becomes the strategy itself, organisations risk losing sight of the outcome they are trying to achieve.
With its extraordinary capabilities and widely publicised potential for competitive advantage, the pressure on organisations to adopt AI is understandable. Yet speed without alignment can leave organisations with siloed pilots and disconnected AI tools rather than a coherent route to long-term transformative value.
McKinsey’s 2025 State of AI research found that AI use has broadened, but nearly two-thirds of respondents said their organisations had not yet begun scaling AI across the enterprise. That finding points to a gap between adoption and infrastructure.
That gap is often organisational. Legal, compliance, security, procurement, and business teams should be aligned while the problem is still being defined, not only when a deployment is waiting for approval.
Before selecting a solution, organisations should map the process they want to improve. Then, they can decide whether AI is actually the right tool rather than a process reform or alternative technologies.
This discipline is especially important in regulated sectors such as healthcare and finance, where AI may touch sensitive data and critical services. These organisations already face intense scrutiny and regulation, raising the stakes for AI use. On the other hand, regulated sectors are also used to operating within rigid regulatory frameworks. With established risk management controls and other governance measures, making the task one of adapting established controls rather than starting from scratch.
Design pilots for real conditions
Pilots have a useful role to play. They test proof of concept and build confidence before wider deployment. This is especially useful within high-risk settings or where user acceptance is in question.
However, while pilots are a preliminary indicator for success, real-world deployment is different. It should also be governed in a way the business can sustain after supplier supervision reduces.
Measures of success should reflect that reality. Productivity gains and time savings are useful, although they are narrow metrics on their own. A tool that saves time but is not trusted by employees will not be used consistently and the ROI will be difficult to gauge. Cultural adoption then becomes part of the equation. Employees are more likely to use AI well if they understand its purpose, its limits, and the role of human judgement. Training and clear messaging that AI is intended to augment work can help avoid blind reliance or adoption resistance.
Pilots are indeed a pathway to adoption, but the true foundation for successful AI transformation is AI Governance, which provides an operating model for the next phase and beyond.
Gartner’s 2025 Hype Cycle for Artificial Intelligence identifies AI agents and AI-ready data as two of the fastest advancing AI innovations. Its analysis points to a wider lesson for businesses: value depends on having the infrastructure, governance, and business alignment needed to support AI beyond isolated pilots.
Build readiness into delivery
AI relies on a constantly shifting data landscape. AI-ready data is not a general state that can be declared once and then assumed for every deployment. Due to this critical reliance and the potential risks from AI use, legal regulation is getting tighter. Platforms that help unify data, trace data provenance, and preserve audit trails can make later governance and compliance easier, particularly where multiple applications or suppliers are involved.
AI system design may involve a combination of LLMs, machine learning, retrieval layers, prompts, data controls, integrations, and user-interface choices that affect performance, accountability, and liability. Accordingly, organisations are increasingly working with lawyers as business partners and documenting design and model choices, data flows, what testing was carried out, and how the system will be monitored after go-live. That record is useful for compliance, but it also helps the business adapt as use expands, commercial strategies evolve, or systems advance.
Make trust part of scale
Supplier selection is a critical pillar in strategic and responsible AI use. A promising use case can become difficult to expand if the supplier cannot support integration, auditability, or changing regulatory requirements. Responsible AI should therefore be part of procurement from the start, rather than a separate review once enthusiasm has built.
Suppliers with experience in regulated environments are often better placed to support this because they are familiar with documentation, privacy, security, and monitoring obligations. That experience can help customers connect legal due diligence with practical delivery.
Businesses may also need to explain how an AI-enabled system operates, including when people are interacting with AI. If that evidence is difficult to understand from the supply chain, scaling the tool may become harder than expected.
The governance gap is already visible. A 2026 UNESCO and Thomson Reuters Foundation report, based on information from around 3,000 companies, found that 44% reported having an AI strategy, while only one in 10 had publicly committed to following an AI governance framework.
AI governance is often framed as a brake on innovation; in practice, it enables more durable adoption in a landscape shaped by rapidly evolving frontier technologies, geopolitical sensitivities, complex technology stacks, and increasing regulation.
Long-term AI value will come from treating the technology as part of business transformation. Organisations that address solid governance and trust before they race to scale will be better placed to move from experimentation to meaningful enterprise impact.
Rebecca Pluthero
