Estimated reading time: 9 minutes

Key points:

  • Many organisations already run a few AI pilots, but without direction. So it stays at scattered experiments that never scale.
  • An AI strategy is a set of choices: which processes first, where AI sits in the organisation, how you help your people learn, and where you want to be in a year.
  • AI is new and has to be owned somewhere. Name who guards the direction and watches the risks and rules, otherwise no one steers it.
  • Free up time to teach your teams to work with AI: what it is, the risks, how it works and how they use it for their own work.
  • It does not have to be a thick report. A course on a single page is enough to steer by and to bring your team along.
  • The strategy comes after your first results, not before. Then you base the choices on what you have seen in practice.
  • A course is not a one-off document. You adjust it each quarter based on what works and what stalls.

Contents

  1. Why scattered pilots stall
  2. What an AI strategy is and is not
  3. The choices that make a course
  4. Give AI a place in the organisation
  5. Make time for your team to learn
  6. A course on a single page
  7. Keep the course alive
  8. How we can help
  9. Frequently asked questions
  10. Sources

Most organisations started with AI long ago. Someone uses ChatGPT for emails, someone else has reports summarised, a team ran a pilot with a chatbot. What is missing is coherence. The pilots stand apart, no one knows what they add up to, and scaling does not happen.

An AI strategy brings direction to that. Not through a thick plan, but through a few clear choices. Below you read what such a course involves, what it contains, and how you build it without months of meetings.

Why scattered pilots stall

Scattered experiments are a good start. They show what AI can do and take away the cold feet. The problem arises when it stops there.

Without direction the pilots compete for the same attention and budget, while no one can say which one delivers the most. What works in one team stays there because no one has the mandate to roll it out more widely. And as soon as the enthusiastic employee who started it moves on to something else, the pilot grinds to a halt. The result is an organisation that is busy with AI and sees little of it come back.

The missing link is a choice about direction. Not more experiments, but a decision about which ones you take seriously and where you are heading.

What an AI strategy is and is not

With "strategy" many people think of a thick document, an external agency and a months-long programme. For AI that is unnecessary and often even harmful, because the field changes too fast to keep a fifty-page plan standing.

An AI strategy is a set of choices that fits on a single page. It sets which processes you tackle first, where AI is owned in the organisation, how you help your people learn, where you want to be in a year, and what limits apply around your data. There are not many choices, and without them everything stays noncommittal.

What it is not: a promise that everything changes at once, or a technical blueprint of which tools you will use. The tool choice follows from the processes you pick, not the other way around.

The choices that make a course

A usable AI course sets out a few things.

  • Which processes first. You cannot do everything at once. Pick two or three processes that cost time, come back often and run on information you already have. Weigh them on impact and feasibility, and start with the processes where the gain shows fastest.

  • Where AI is owned. AI is new and needs a fixed place somewhere, otherwise no one steers it. Name who guards the direction and watches the risks and rules, and give each chosen process an owner who drives the rollout. I work this out further below.

  • How you help your people learn. Set aside time to teach your teams to work with AI: what it is, the risks, how it works and how they use it for their own work. I work this out below too.

  • Where you want to be in a year. A concrete picture: which processes run on AI by then, and what that has delivered in time or quality. To decide what a realistic next step is, it helps to know where you stand now. A maturity model maps your current level, from scattered pilots to AI that sits firmly in your processes.

  • Where the limits lie. Which data may and may not go into an external tool, and what agreements apply with customers. As soon as personal data goes in, the GDPR and your own agreements apply. This does not have to be a full policy, a few clear rules are enough to start.

Wireframe network of nodes with one larger anchor node and turquoise pulse lines, representing AI owned by one place in the organisation

Give AI a place in the organisation

AI is new, and new things get left undone if no one manages them. An AI strategy therefore names where AI belongs in your organisation: one person or a small team that guards the direction, makes the calls on which tools and processes are in and out, watches the risks and the rules, and where colleagues can go with their questions.

It does not have to be a new department. Often it is an existing manager with AI in their portfolio, helped by a few people from different teams. The point is that it is owned somewhere, so decisions do not stall and risks are spotted in time. Without that place AI stays something everyone does a bit and no one steers.

Make time for your team to learn

A course only succeeds if the people who work with it know what they are doing. So there has to be fixed time for learning in your strategy.

It comes down to a few things your team gets to grips with: what AI is and what it cannot do, what risks come with it such as wrong answers or data you do not want to share, roughly how it works behind the scenes so they understand its limits, and how they make it work for their own job. Those who get this use AI with more confidence and make fewer mistakes.

Set aside real time and budget for it; expecting people to fit it in on the side does not work. How you approach that learning and the broader adoption, you read in getting your people on board: AI adoption in your team.

A course on a single page

The good thing about these choices is that together they fit on a single page. That is not simplification for its own sake, it forces you to choose. What does not fit on the page, you do not do this year.

A workable layout: at the top the one-year goal in two sentences. Below it the two or three processes you tackle first, each with its owner, and who guards the whole. Next to that the limits around data and the time you free up for learning, in a few lines. And at the bottom how and when you discuss progress.

That one page is enough to steer by and, just as important, to bring your team along. In a few minutes everyone can see where the organisation wants to go with AI and what that means for their own work. No one reads a thick report; a clear one-pager hangs on the wall.

Wireframe closed-loop cycle of four stations with a turquoise feedback stream, representing adjusting the AI course each quarter

Keep the course alive

An AI strategy is not a document you make once and then put away. The field changes, you learn from the first processes, and what was impossible a year ago is possible now.

So plan a fixed moment, for example each quarter, to hold the course up to the light. Which processes run as intended, which lag behind, and what is the next one you pick up. That way the strategy keeps moving with what you see in practice, instead of ageing in a drawer. So the first version does not have to be perfect; you adjust it along the way.

How we can help

If you want to get your management team aligned in a short time and draw up that one-page course together, the workshop AI Strategy for Managers is the starting point. You go home with the first processes chosen, the owners named and a goal for the coming year.

If you then want to carry that course out as well, from choice to working application, we help with the delivery in AI in Business. We take on the chosen processes, set them up inside your own environment, and make sure progress stays measurable.

Book a no-obligation call, and we look together at where the first gain lies for your organisation.

Frequently asked questions

Do you need an AI strategy?

As soon as you want more than scattered pilots, yes. A strategy makes choices about which processes first, who does what and where you are heading. A course on a single page is enough to begin.

What goes into an AI strategy?

Which processes you tackle first, where AI is owned in your organisation, how you help your people learn to work with it, where you want to be in a year, and what limits apply around your data. Together they fit on a single page.

How long does it take to make an AI strategy?

Not months. If you already have some experience with a first process, you draw up the first course in a few sessions. More important than the length is that you then adjust it each quarter.

Does the strategy have to be there before we start?

No. Start with one process and book a result there, and build the strategy on what you learn from it. That way you build the course on what works in practice, not on assumptions up front.

Who makes the AI strategy?

Management, because it is about choices on direction, people and budget. You do not need a technical specialist for it; the tool choice follows later from the processes you pick.

Who is responsible for AI in a company?

Put it with one person or a small team that guards the direction, watches the risks and rules, and where colleagues can turn. It does not have to be a new department; often it is an existing manager with AI in their portfolio. What matters is that it is owned somewhere, otherwise decisions stall.

Should you train your team in AI?

Yes. Set aside time to teach your people what AI is, what risks come with it, how it works and how they use it for their job. That belongs as a fixed part of your strategy. More on that in getting your people on board: AI adoption in your team.

Sources