AI adoption among small and medium-sized European companies remains stubbornly uneven (only 14% of small businesses use AI today), compared with 34% of larger ones. Behind that gap, most leaders are facing five barriers at once.
AI tools cost real money before they save any. Tight margins mean tools must earn their place before they go in, and most procurement processes weren't built for that.
Businesses don't have the tech experts. Internal teams are not digitally savy enough. The research and case studies that exist are not focusing on their specific needs.
Legacy IT doesn't play well with AI, leading to expensive integration issues. Data silos make it more difficult.
People are skeptical. Some fear for their jobs; some have seen IT projects fail before. Without leadership clarity and proper training, adoption stalls regardless of how good the technology is.
ROI on AI can take two years or more — longer than most SME planning horizons. And GDPR, the EU AI Act, and sector-specific rules add another layer of uncertainty about what's even allowed.
Most AI engagements assume the basics are in place: clean data, modern systems, a clear strategy.
Unfortunately, that's rarely the case.
I work with companies where the question isn't which AI model
but where do we even start.
The answer almost always involves untangling the data, the systems, and the strategy at the same time.
The management, the team, the competitors, the vendors, all have an opinion about AI. They often disagree. It needs someone who can sit and listen to the different players, actors, stakeholders and tell what's worth doing, and what isn't
Is is not rare to have systems that give different numbers, data siloed or sitting in spreadsheets nobody owns. There's no shared definition of what a customer, a sale, or a product even is. Before AI can do anything useful, this needs to be sorted, and it's solvable.
Years of tactical decisions can translate in multiple systems which, sometimes, don't talk to each other. Replacing one risks breaking three others. Adding AI on top is impossible without first deciding what stays, what goes, and what connects to what.
Every engagement follows the same shape — discover the work, then deliver the value — with a two-week proof of concept as a low-risk on-ramp.
Seven phases of discovery and delivery — with an optional two-week proof of concept as the on-ramp.
Forty actions across eight steps — the full toolbox for a process and data engagement.
All of it executed under EU regulatory discipline — GDPR and the EU AI Act as the source of authority, EU AI Act risk tiering and the DCAM data capability model as the conceptual frames, the Carlo toolbox as what gets delivered.
Identify and prioritise the AI opportunities most likely to move company's numbers and prove it with disciplined pilots that scale.
Diagnose data foundations and build the operating model that unlocks AI, without a two-year platform rebuild first.
Stand alongside executive teams as an interim or fractional advisor, taking accountability for outcomes through delivery.
You pick one target process. We design, build, and validate a working AI solution in two weeks. You see exactly what's possible, at low risk. Then you decide: scale it, iterate, or explore something else.
We work best when the target process has been identified, with messy data or manual workflows, and a team ready to move fast. If you think you have this, let's demonstrate the benefits.
A 30-minute conversation. You describe your challenges and I explain how I can help.
