I recently wrote that there is no such thing as an AI transformation. I still believe it. What exists is business transformation that uses AI as one mechanism among others, based on a clear definition of organisational purpose and strategy. This white paper is the other half of that argument.
The paper builds on papers produced by Ilana Sprongi and on work she and I are doing jointly for the Agile Business Consortium.
Why so many AI initiatives are being abandoned
S&P Global Market Intelligence found that 42 per cent of companies abandoned most of their AI initiatives in 2025, up from 17 per cent the year before. The cause is rarely the algorithm. The pilot was never connected to a business outcome. Too many organisations have AI solutions looking for business problems.
I saw the same mistake with Agile. Transformations stalled because they promised business benefits but treated change as something that happened in software teams. AI transformation is repeating the pattern.
What AI needs to earn its place
In the paper, I argue that AI earns its place in a transformation when five things are true. It is tied to a specific strategic goal. It is defined as a solution with a business case. It has been assessed for the harm it could do. It is bounded by operational controls. And it has a way to stop it if it starts to misbehave. Leave out any one of those and you have an experiment, not a transformational capability.
What the paper covers
The paper starts with the strategy map rather than the use-case list, because an AI opportunity is only ever right or wrong relative to a specific strategy. It then looks at what leaders should hold constant and what they should adapt, including a client example where an engineer-to-enabler ratio triggered an organisational redesign.
The second half sets out four linked artefacts from my AI Authority and Governance Toolkit: the business case, the risk assessment, the controls and the stop-use mechanisms. It covers Threshold Matrices, organisational drag, autonomy as a setting rather than a switch, kill switches and reversion models. It closes with how to keep governance honest after go-live, including what Provision 29 of the UK Corporate Governance Code means for evidence that your controls work.
Most AI initiatives do not fail in the data centre. They fail on a Tuesday afternoon in a steering committee where nobody is sure who owns the decision the model is now making. This paper is about making sure someone does.
Download the full whitepaper:
The Truth About AI and Transformation: download the PDF
If you have an AI initiative you are not yet sure you could stop, I would like to hear about it. Get in touch.
Related reading: Why Your AI Roadmap Should Start With a Strategy Map, Not a Use-Case List and Strategy Mapping: Using AI to Do the Legwork.
