AI strategy for business: from experiments to an implementation plan

My thesis based on conversations, experience, and work with clients

In my experience, the problem most companies face today is not a lack of ideas about AI. The problem is rather an excess of ideas, tools, and expectations that don't add up to any coherent whole. In this situation, organizations don't need another demo or another presentation about trends. They need an AI strategy — a calm answer to the question: where does AI make sense for us, and where does it not yet.

What this means in practice

In practice, it usually looks quite similar. Somewhere in the organization, the first experiments appear. Someone tests ChatGPT, someone else talks to a vendor, yet another person prepares a proof-of-concept. From the board's perspective, it looks like movement, but it is often only apparent movement. The more scattered initiatives there are, the harder it becomes to answer what the company really wants to achieve and who should be responsible for it.

That is why an AI strategy should not start with a question about tools. It should start with a question about business priorities, organizational readiness, and what type of implementation is appropriate today. Sometimes it will be a pilot. Sometimes a workshop. Sometimes governance. And sometimes the decision that now is not the right time.

Why this is a problem right now

The pressure around AI is growing faster than organizational maturity. Boards want to see results, teams want to experiment, vendors make bold promises, and in the middle sits the CIO, CTO, or innovation leader who has to make sense of it all — and still explain why not every idea is worth implementing.

This is exactly where an AI strategy becomes necessary. Not to slow things down, but to avoid confusing activity with progress.

What actually works

What actually works is usually less flashy than what looks best at conferences. What works is a clear connection between AI and a specific business problem. What works is sensible selection of use cases. What works is an honest assessment of data, processes, and risks. What also works is assigning ownership to decisions. Without that, even the most promising projects quickly turn into costly experiments.

That is why a good AI strategy is not a catalogue of ideas. It is rather a filter through which only those initiatives pass that make business, organizational, and operational sense. It is precisely this filter that most often distinguishes companies that are actually building something with AI from those that are merely circling around it.

How I work on this with clients

I usually start by talking to the people who actually carry this topic on their shoulders: business, technology, and the people responsible for operations. I am interested not only in what the organization wants to do, but above all in why it considers this important right now and what it has already tried.

Only then do we move to organizing use cases, constraints, and possible paths forward. Sometimes the result is a roadmap for several months. Sometimes a decision to conduct a readiness audit. Sometimes a workshop for the board. Sometimes a significant narrowing of scope. And that too is a good decision, because a strategy should not produce movement. It should help choose a sensible direction.

My conclusion for Boards and CTOs

Don't ask first: "What AI tool should we buy?". Ask instead: "What problem do we want to solve, what change do we expect, and is our organization ready for this move?". Because only then does the conversation about AI stop being a conversation about fashion and starts being a conversation about decisions.

FAQ

How does this service differ from an AI readiness audit?

An AI readiness audit primarily answers the question of where the organization stands today. An AI strategy goes further and helps decide what direction of action makes sense and what should happen first.

Is AI strategy for companies that haven't done anything with AI yet?

Yes, but in that case, the best first step is often not a full strategy right away, but rather a shorter audit or discovery workshop. A strategy delivers the most value where decisions need to be made, not just reconnaissance.

Do you select specific tools as part of this service?

Yes, but only once it is clear what they are meant to serve. A tool is a consequence of strategy, not its starting point.

Is this a service only for large corporations?

No. The scale of the organization changes the scope and number of stakeholders, but the need to structure decisions is similar in both larger and smaller companies.

Get in touch

Dear Reader. If you see that the topic of AI strategy is relevant to your company today and you would like to have a calm conversation about where to start, I invite you to get in touch. Not to immediately launch a large program, but to sensibly set the first or next step.

For editorial review

• Do we want to explicitly promise a 90-180 day roadmap as a standard deliverable on this page?
• Do we reference the AI Act and governance here, or do we leave that clearly to the dedicated page?
• Do we add a "when not to do an AI strategy" section to more strongly filter out the wrong leads?