The sentence that ends most AI conversations

A CEO I work with told me early on that he did not want to hear anything about AI until it was actually implemented in his company.

Not rude. Just tired. He had sat through the presentations, read the posts, heard the promises. None of it had changed a single thing about his Monday morning.

He was right to be tired. And I agreed with him, which surprised him a little.

1 / What actually changed his mind

A few weeks in, he spoke a brief out loud to ChatGPT. Find everything a colleague had sent about a trade fair happening in early 2027. Archive it in a shared folder. Write me a summary of the deadlines. Set up a project for it and invite the two people who own it.

2 / Then it just did it

Twenty-four emails, sixteen threads, one folder, one brief, one project with the right people invited. He sat quiet for a second and said it was completely insane. Then he said something better: okay, I just have to check it. That makes sense.

3 / He stopped being an operator

You could watch the shift happen on his face. He was not learning a tool. He was realising that his job had quietly changed shape. From now on he describes the outcome and checks the result. He orchestrates, and ChatGPT does the moving.

4 / Which means the micro-tasks are finished

Jumping between tabs. Clicking through folders. Downloading a file to upload it somewhere else. Copying text out of one window and pasting it into the next. None of that was ever his work, it was the tax he paid to get to his work. For him and for his team, that tax is now optional.

5 / Sample data convinces nobody

I have demoed AI on invented examples many times. It never lands. The moment the material is their own mess, with their own colleague's name on it and a deadline they actually fear, the argument is over. There is nothing left to debate.

6 / The output was only as good as the brief

This is the part people skip. He is a precise person who delegates well to humans, so he delegated well to a machine. He said what he wanted, in what order, for whom. Vague people get vague results and then blame the model. The skill being tested here is not technical.

7 / Save the thing that worked

When a chain of steps runs end to end, save it. Most tools now let you keep a working process as something you can call again, with all the corrections you made along the way already baked in. Do that once and you stop re-explaining yourself every week. Skip it and you are starting from zero forever.

The skeptics are not the problem. They are usually the most careful people in the building. They have been told for three years that everything is about to change, and every Monday has looked the same, so they have learned to wait for proof instead of promises. That is a fair thing to ask for.

My next step with this company is the whole organisation. I am starting at the top on purpose. A convinced CEO raises the flag and the rest of the team follows, and nothing I could say in a training session carries the weight of their own boss saying he has seen it work.

Talk soon,
Mikkel

P.S. If you had to pick one task in your business to do live in front of your biggest skeptic, which would it be? That is probably the one to start with.