Pedro Olivares
Building with Schneider Electric Across Europe

Schneider Electric is present in more countries than most companies have offices. That is the first thing you learn when you start working with them, and it changes everything about how AI gets built.
We are working with Schneider Electric to take AI use cases into production across several European countries at once. This post is not about the use cases themselves. It is about what “across countries” turns out to mean, and how we are approaching it.
One company, many realities
From the outside, a multinational looks like one company. From the inside, it is a federation of local realities. The same process runs on a different system in Spain than in Germany. The same document is a PDF in one country and an email thread in another. The same customer question arrives in six languages and is answered by six teams with six sets of habits. None of this is dysfunction. It is what decades of growing across borders produce.
Most AI projects treat this as a problem to solve before they start: harmonise first, automate later. That project never ends. We treat it as the terrain. The agent has to work in the reality that exists, in each place, from the first day.
Start where it hurts, in one place
We do not begin with a European rollout. We begin with one country, one team, and one thing that is slow, repetitive or expensive enough that people are willing to change how they do it. We follow the same discovery we always use: listen before challenging, and choose use cases for their odds of shipping rather than their odds of impressing. When it works, it works in one place, with real people relying on it every day. That is the only kind of success worth repeating.
Then repeat without copying
This is where the approach differs from a classic rollout. What travels from the first country to the second is not the agent. It is the pattern: what the agent does, what it is allowed to do, how it is checked, what it connects to. The second country receives the pattern and makes it its own: its language, its systems, its rules. The core is shared; the edges are local. It is closer to how Schneider Electric already runs than to how software is usually deployed.
Because everything lives on one platform, the second implementation takes a fraction of the first. The third takes less again. Local teams do more of it each time, and by the fourth country we are mostly watching.
Europe is a constraint, and an advantage
Working across Europe means working across regulators, works councils, data residency requirements and languages. We treat that as a design input, not a hurdle. Everything runs in Europe, data stays where each country needs it to stay, and every decision an agent takes is recorded so that it can be explained later, to anyone who asks. Talking to an agent in Italian, Polish or Spanish is not a feature we add at the end. It is where we start.
The advantage is less obvious but real. A company that has learned to operate under many rules at once already has the muscle that AI in production requires: clear ownership, documented processes, and the habit of asking “who is accountable for this?” before “how fast can it go?”. Those are exactly the questions that make an agent trustworthy.
What it looks like when we leave
The measure of success is not the number of countries. It is how many of them run without us. Our engineers build the first one with the local team, not for them. By the time a use case has crossed two or three borders, it is Schneider Electric’s own people proposing the next country, and the next use case, from inside their own studio.
That is the plan: a company that already knows how to be in many places at once, with AI that can keep up.
We will share more as the work matures. In the meantime, if you run a business in more than one country and the word “rollout” makes you tired, talk to us.


