Why AI consulting on its own won't get you there.

1. The problem isn't the technology.

Most AI projects don't fail because someone picked the wrong model. They fail because the consultants leave and nobody's left to build the thing. The consultant hands over a roadmap. The company nods along. Then reality kicks in: support tickets, holiday cover, the day job. The AI work slides down the list. Eventually nobody asks about it at all.

2. Workshop days without working output are expensive inspiration.

I know the format. Two days, good energy, whiteboards full of Post-its. Then a handbook. Then silence. The workshop itself isn't the problem. The problem is handing over a slide deck instead of a system. Every workshop I run ends with something that actually works the following Monday. No roadmap. No homework. A result.

3. If you haven't built it yourself, you don't know what works.

Consulting without hands-on experience is theory. You can map processes, compare vendors, draw decision matrices, and still have no idea how an automation behaves when the API times out at 3am. Companies only find that out once they start building.

4. GDPR isn't an excuse - it's a requirement.

"We can't do that because of GDPR" is the thing I hear most before a project starts. It's usually wrong. Powerful, compliant systems exist: self-hosted, on German servers, with proper data processing agreements in place. Anyone using GDPR as a blocker hasn't looked closely enough. Or doesn't want to start.

5. AI changes nothing - until it's running.

Strategy matters. Strategy without a working system is wishful thinking. Value shows up the moment the first process runs on its own, the first hour is saved, the first colleague picks up the new tool and won't go back. Everything before that is preparation. Necessary preparation - but not a result.