Anna Totterdell
Projects Director
There is an industry being built around your confusion about AI, and it is not in your interest.
Over the past two years, a new category of consultancy has emerged: the AI strategy firm. They come in various shapes - boutique consultancies, big four practices, freelance "AI transformation" specialists - but they all sell essentially the same thing: an assessment of your AI readiness, a roadmap of opportunities, and a set of recommendations that invariably require more consulting to implement.
This is not a conspiracy. It is a business model. And it is failing you.
The AI strategy trap
Here is how it typically works. You engage a firm to assess your AI opportunities. They run workshops. They interview stakeholders. They review your technology stack. After six to twelve weeks, they deliver a document - sometimes quite impressive, often beautifully designed - that outlines where AI could add value in your business.
The document contains phrases like "significant opportunity," "potential for automation," "AI-enabled decision-making," and "digital transformation roadmap." It identifies a dozen or more use cases. It recommends a phased approach. And it concludes with a proposal for the next phase of work: building a proof of concept.
You spend the money. You get the proof of concept. It works in a controlled environment. And then nothing happens.
Nothing happens because the proof of concept was built in isolation, on clean data, outside your actual workflows. Moving it into production would require data and systems integration with your existing platforms, cleaning your data, redesigning your processes, and building operational governance around the AI output. None of which was in scope, because none of it was part of the strategy.
And so you are back where you started - except now you have spent fifty thousand pounds and six months, and your team is more cynical about AI than they were before.
Why this model is structurally broken
The problem is not that these firms are incompetent. Many of them employ genuinely talented people who understand AI deeply. The problem is that the model separates strategy from execution in a domain where that separation is fatal.
AI does not work in theory. It works in operations. It works when it is embedded in a specific workflow, processing specific data, producing specific outputs that feed into specific decisions. The only way to know if AI will work in your business is to connect it to your business - your actual data, your actual processes, your actual systems.
A strategy document cannot tell you that. A proof of concept on synthetic data cannot tell you that. The only thing that tells you that is implementation - real, messy, operational implementation.
And most AI consultancies do not do implementation. They do strategy. They do roadmaps. They do innovation theatre. And then they hand you a document and wish you luck.
What you actually need
You do not need an AI strategy. You need AI enablement - someone who will look at your operations, identify the specific points where AI can add measurable value, and then build it inside your systems, on your data, connected to your workflows.
That means the people advising you need to understand three things: your operations (not just your technology), your data (not just your tools), and your capacity (not just your ambition).
They need to be able to tell you: this specific process, which currently takes your team fourteen hours a week and involves manual reconciliation between three systems, can be reduced to two hours by automating the data extraction, normalising the inputs, and using AI to categorise the exceptions. Here is how we build it. Here is what it costs. Here is when you will see results.
That is not a strategy. That is a scoped, deliverable, measurable project with a clear outcome. And it is the only kind of AI investment that reliably pays off.
The hard truth about AI value
AI creates value in exactly one way: by changing an operational metric. Reducing cost. Increasing speed. Improving accuracy. Eliminating manual work. If you cannot point to a specific metric that AI has improved, you have not created value. You have created a demo.
The businesses that are genuinely benefiting from AI right now are not the ones with the most sophisticated strategy. They are the ones that skipped the strategy phase entirely and went straight to: what is the most painful, manual, error-prone process in our business, and can we fix it?
They started small. They started with data. They started with operations. And they built from there - iteratively, measurably, pragmatically.
Stop buying strategies. Start buying outcomes.
If someone offers you an AI strategy, ask them one question: what metric will be different in twelve weeks? If they cannot answer that - specifically, with a number - they are selling you a document, not a result.
The AI industry wants you confused. Confusion creates demand for advice. Advice creates demand for more advice. And the cycle continues, while your operations remain unchanged and your competitors quietly get on with the actual work of embedding intelligence into how they run.
You do not need more advice. You need tech consultancy that actually makes things work. That is a fundamentally different proposition, and it is the one you should be buying.
