The Border Is Part of the Job

This post is part of my Medium blog.

The first thing you learn about a multinational company is that the org chart lies about distance.

The box says Vancouver. The other box says San Jose. The line between them looks clean in the diagram, which is how you know the diagram was not made by anyone who has actually traveled between the boxes with a laptop, a passport, and a meeting that starts in three hours.

In Redundant, Rob Coleman lives in Vancouver. The company has offices in the United States, and the first major meeting in the book requires him to fly south to San Jose. Officially, Rob has a meeting. The real version is that the border is part of his job.

Crossing a border for work was never frictionless. Usually it's fine — you answer the questions, show the documents, and continue. But "usually fine" is not the same as easy, and it is not getting easier. The questions are getting more specific. The scrutiny is getting less casual. What used to be a routine crossing now carries a real chance of an interview you did not plan for.

The same friction shows up in the budget. A company wants one global view of the numbers, which usually means converting everything into U.S. dollars. Rob's team lives with Canadian salaries, Canadian taxes, and Canadian housing costs. The model wants one currency. The people do not live in one currency.

None of this is exotic. It is the ordinary machinery of working across a border. What makes it interesting now is AI.

Companies want one platform. They want to move data, models, services, and people across borders the way they move invoices. But governments are not converging on what AI risk means. One country treats a model as a product with a safety case. Another treats it as infrastructure. A third has not decided yet. The definitions of risk, accountability, and acceptable use are being written jurisdiction by jurisdiction — and they are not coming out the same.

That matters because AI work is border work now. Where is the training data allowed to come from? Where can the model be served? Who is liable when it is used in a country with different rules than the one it was built in? The model does not care where it runs. The regulators do.

In the first book, Rob is dealing with the ordinary version: a FinOps team spread across countries, budgets that have to be converted, and a meeting that requires him to cross a border before he understands why he was invited.

The later books widen the frame. The relationship between Canada and the United States becomes more complicated.

I'm not going to explain how that develops.

The point is that the border was already in the room before anyone started talking about AI regulation. It was in the travel schedule and the payroll model. Regulation adds a new set of borders on top of the old ones — and this time the crossing is not only about people.

The company is one organization on paper.

The people still have to cross. Subscribe to The Condition Set newsletter