Adoption is accelerating faster than organisational clarity

AI has moved from experimentation into mainstream management attention. Stanford's 2025 AI Index reported that 78 per cent of organisations said they used AI in 2024, up from 55 per cent the year before. OECD research on AI adoption in firms describes skills, data, finance and organisational capabilities as important complements to the technology. The implication is clear: access to tools does not create enterprise capability by itself.

Microsoft's 2025 Work Trend Index, based in part on a survey of 31,000 workers across 31 countries, found that 82 per cent of leaders saw 2025 as a pivotal year to rethink strategy and operations. Vendor research should be read with its commercial context in mind, but the scale and direction of the finding align with a wider executive reality. Leaders are being asked to make operating-model choices while the technology, regulation and workforce response continue to change.

A training-first response is too narrow. It can teach leaders what a model can do without helping them decide where AI should be used, who owns the outcome, what evidence supports the investment or how risk will be controlled.

Five capabilities leaders need

AI capability begins with management judgement. Leaders do not all need to become technical specialists, but the executive group needs enough shared understanding to govern choices and reshape work.

  • Value framing: identify business problems where AI can improve a measurable outcome, rather than collecting disconnected use cases.
  • Work redesign: decide how tasks, roles, controls and customer interactions change when human and machine capabilities are combined.
  • Decision rights: clarify who approves use cases, data access, models, deployment and exceptions.
  • Risk judgement: understand model limitations, privacy, security, bias, explainability and the consequences of failure in context.
  • Adoption leadership: create the incentives, learning loops and psychological safety required for responsible use at scale.

Move from use cases to an operating system

Many organisations begin with pilots. This is useful for learning, but a portfolio of pilots can become another form of activity without control. Leaders need a mechanism for selecting opportunities, testing value, assigning risk ownership, redesigning processes and stopping work that does not meet the threshold for scale.

The operating model should connect strategy, technology, data, risk, workforce and change. It must define how ideas enter the portfolio, what evidence is required at each stage, how reusable capabilities are built and which decisions remain human. It should also give employees a credible account of what will change, what will not and how the organisation will support new ways of working.

The strongest first step is often a leadership capability session linked to real decisions. Instead of generic awareness, the executive team works through its own value cases, governance choices, workforce implications and risk appetite. This creates a common language and exposes the decisions that must be made before experimentation can become managed transformation.

Sources and further reading

Rooted's interpretation combines these published sources with practical transformation experience. Statistics and research findings are attributed to their original publishers.

All insights