Why AI Adoption Is Not an IT Project

AI adoption is change management: why almost everyone gets it wrong

There's a reason most AI projects in SMEs never deliver visible results. AI adoption requires change management, and only 28% of German companies actually have that in place. The rest treat AI like a software rollout, then wonder why nobody uses the tools.

IT is already stretched thin. Yet the AI initiative lands there anyway. New logins get set up, licences get handed out, an email goes round to everyone. Then everyone waits. Employees try it once, find it underwhelming, and go back to their old way of doing things. A quarter later, the board asks what it actually achieved.

Nothing. Because technology on its own doesn't change behaviour.

What "change management is missing" actually means

According to Mittelstand Digital's 2025 AI Study, 72% of SMEs have no change management process for AI at all. No communication explaining why it's happening. No involvement from the people affected. No support through the transition. No feedback channel.

That's not a minor detail. It's the reason 67% of employees have reservations about AI, according to the HKA study.¹ When you introduce new technology without explanation, without purpose, and without involving people, they respond with scepticism. Not because they're against technology. Because nobody did the obvious thing: told them what it means for them.

Change management sounds like management consultancy and PowerPoint slides. It isn't. For AI, it means employees know what's happening. They can contribute ideas. They see themselves in the new setup, not as casualties of an automation wave.

That's the difference between AI that gets used and AI that just sits there while everyone pretends otherwise.

Why IT can't solve this problem

AI too often lands with IT. And IT is already stretched to the limit.

Worse: IT doesn't understand the business problems well enough to find the right use cases. It can judge what's technically feasible. But it doesn't know what costs the sales team three hours every Monday. It doesn't know which step in quality control drives everyone crazy. It doesn't know why accounting reprocesses the same thing by hand every month.

AI projects managed solely by IT solve IT problems. Not business problems.

Treat AI as an IT rollout and you'll get IT results: infrastructure that works, tools nobody uses, and no measurable change to daily work.

AI exposes what was already broken

AI acts like a magnifying glass over everything that's not working in an organisation.

Workflows nobody has questioned in years suddenly become visible the moment you introduce AI. Processes that "have always worked this way" turn out to be built from a decade of habits that grew organically, with no real design behind them. If AI is going to take over a process, that process has to be clear first. Often the first real finding is: we don't need AI here. We need order first.

That moment isn't a setback. It's an opportunity. AI adoption is always an organisational stock-take too. See it that way, and you win twice.

A bad way of working stays bad whether a human or an AI runs it. AI just makes it go wrong faster.

Who actually needs to be in the room

Effective AI adoption needs four perspectives in the room at once:

1. Process knowledge: Who does the work? Who knows the pitfalls of daily operations? That's the people on the shop floor, in sales, in customer service, not management three levels up.

2. Technical judgement: What can realistically be built with the data and infrastructure you already have? What costs too much? What's doable in weeks instead of years?

3. Business perspective: Which problem gives you the most leverage? Which one, once solved, drags several other problems along with it? Where can you measure added value instead of just describing it?

4. Transition support: How will the changes be explained? Who guides the team through the transition? Who's the point of contact when people are unsure?

Miss one of these four and the project fails. Not because of the model, not because of the data, but because of the process.

AI projects decided purely top-down produce results nobody uses. That's not bad luck. That's predictable.

AI adoption with change management: the people project

AI is an organisational change. It's not a new machine you switch on and leave running. It's not a software licence that shows ROI after 90 days. The first few months are R&D: understanding processes, rebuilding workflows. None of that shows up on the P&L. But it lays the foundation for everything that comes after.

Companies that get AI adoption right treat it as a people project: real challenges from the team as the starting point, real involvement in building the solution, real results the team helped create.

The AI Design Sprint is built around this principle. C-level, IT, business departments and sales sit at the same table. The problems come from the team. So do the solutions. That builds buy-in, because the people who'll use the AI later helped build it. No vendor lock-in, no dependence on a consultant afterwards.

I explain how this fits into a broader AI strategy in my article on AI strategy for SMEs.

The one question to ask before your next AI budget

Before you buy the next tool, ask this: who's actually going to be affected by this initiative, and have we asked them yet?

If the answer is "no", or "after launch", you don't have a technology problem. You have a leadership problem.

Your next step

I help companies design AI adoption so the results actually get used. Not with an IT ticket, but with the right process and the right people in the room. Get in touch if you want to know what that could look like for your company.

¹ Mittelstand Digital AI Study 2025 and Karlsruhe University of Applied Sciences (HKA/KARL Study 2025): 28% have change management in place for AI; 67% report reservations among employees. mittelstand-digital.de and h-ka.de

Back to overview