Why you do not need an AI strategy

I did my first AI project back in 2018. Not because we wanted to “do AI,” but because we had a very concrete business problem.

The objective was to improve the intake of tickets in our customer service organization. Using more than 200,000 messages and their classifications as training data, we were able to recognize language and business case with more than 90% accuracy for new inbound messages. Tickets were automatically routed to the people who could handle the case and spoke the right language.

The reason was straightforward. Team leaders manually assigned tickets to their team members. In my view, that is not what a team leader in a support organization is paid to do. Their job is quality, training, and people development-not acting as a human routing engine.

So we automated the process.

AI - in this case, machine learning - was simply the most effective tool available. Without it, we would not have reached this level of speed and accuracy in such a short time. But AI was not the starting point.

The starting point was a far less fashionable question:
How do we automate this process?

Automation has two purposes. First, stop wasting human brains on work that machines can do better. Second, improve service quality by responding faster and more accurately. That increases customer satisfaction and reduces workload - because customers do not contact support for entertainment.

This boils down to two business imperatives: use people effectively and increase speed and accuracy in customer interactions.

We knew where we wanted to go. We knew why. AI was an enabler - not the objective.

And this is where most “AI strategies” go wrong.

An AI strategy puts technology first. And that is already a red flag. “AI” is a moving target, and a technology-first perspective (on its own) is fundamentally disconnected from how businesses actually create value.

If you look at your organization and ask, “Where can we use AI?”, you will find countless use cases. AI can always do something. And if you calculate ROI for each of them, you will almost always find a positive number. Congratulations - you now have a long list of AI initiatives.

What you probably do not have is a coherent direction.

You optimize parts instead of questioning the whole.

You use AI to extract data from paper documents instead of asking why you still accept paper documents at all. You speed up order intake instead of questioning whether the order intake process itself should exist in its current form - or whether direct integration with partners would bring the data directly into your system.

When you put AI - or any other technology - at the center, it becomes the answer by default. And once you have the answer, you stop asking the uncomfortable questions. You stop challenging how your business actually works.

That is why you do not need an AI strategy as your starting point.

You need an automation strategy. One that starts with a hard look at how your organization works today, where that way of working is slowing you down, and where entirely different approaches could support your business objectives - or open new opportunities.

Only then does the toolset matter. And yes, that toolset may include AI. But sometimes the most effective solution is not a smarter algorithm - it is a different process.

Here is the part many people get wrong: this is not an argument against technology. The opposite is the case:

AI is an enabler. And you will only see enabling opportunities if you understand the tools. That requires real capability building: training, experimentation, and hands-on learning. Not “prompt engineering for everyone,” but the ability across the organization to question how work is done-and the technical literacy to recognize when AI actually helps, and when it is just a shiny distraction.

If your AI strategy comes before those questions, you already started in the wrong place.