TROi

AI Transparency

Short version: the thinking is mine, the typing had help. Since the paper spends a chapter on what AI does and doesn't do for an organization, it would be a bit rich not to say how I used it here.

What I actually did

The observations in this paper come from about twenty-five years of working inside organizations and noticing the same things happening. I carried most of it around unwritten for a long time. Getting it out of my head and into something anyone else could read is the part that took until 2026.

I wrote with Claude, and it earned its place. It argued with me, which was the useful bit. It caught the places where I had said the same thing twice in different words, asked what I meant when I hadn't decided yet, and found the paragraph where I contradicted something I had written eleven chapters earlier. A good editor does that. This one is available at two in the morning.

What it did not do is have the ideas. It has never sat in a steering group watching a decision take three weeks, or run a function, or been wrong about an architecture in a way that cost somebody money. Every observation here is one I have made myself, and every judgement about what to keep, cut, soften or stand behind is mine.

There is a version of this note that lists tools and process steps and reassures you about editorial control. I would rather just say it plainly: the ideas are mine, some of the sentences are better than they would have been on my own, and I read every one of them before it went up.

Christoffer Råsten · TROi · christoffer.rasten@troi.se