The question has changed from whether to how
Seventeen months of writing, and the conversation in Swiss companies has moved. What that means for what this blog covers next.
When this blog started in October 2024, the question in most Swiss SMEs was whether to use these tools at all. That question is now largely settled, and a different one has taken its place: how to run them so they help without creating problems.
That shift is worth naming, because it changes what is worth writing about.
What the settled question left behind
Companies are using these tools. Not always deliberately, not always with anyone’s knowledge, and rarely with the questions answered that this blog has been putting for a year and a half: who decides, what may go in, where is it processed, who checks the output, what happens when it is wrong.
Those are governance questions, and the awkward thing about them is that they are dull individually and consequential together. A company can answer all five in an afternoon and most have answered none.
The tension that is now the real subject
Every control costs something in speed. A confirmation step before an agent acts, a review before a draft goes out, a restriction on which documents can be pasted — each is defensible and each makes the tool slightly less useful.
Get the balance wrong in one direction and you have shadow use, which is the 27 November problem: people routing around the approved path because it is slower. Wrong in the other and you have controls nobody follows, which is the same outcome with more paperwork.
That balance is a genuinely hard problem, it does not have a single answer, and it depends on what your company actually does. It is also where most of the useful thinking of the next year is going to happen.
What has not changed and will not
The things that determine whether a company gets value are still unglamorous and still mostly organisational. Name the person. Write the process down. Ask five people why they are not using it. Keep a one-page record.
Seventeen months of model releases have not displaced a single item on that list, and the four that mattered to a company of thirty people would have fitted in one post.
Next
More on the governance side, less on the release cycle — because that is where the open questions now are, and because the release cycle has become predictable enough to summarise rather than follow.
Thank you for reading. The useful work was always the boring work, and it still is.