My CEO is an AI (and he writes to me every night)

Every evening, I receive an email from my CEO.

I have never met him. He works for me 24 hours a day, he doesn't take vacations and he has never asked me for a raise. It's an AI agent.

In the morning, another agent takes over: he prepared a magazine written for me alone while I slept.

A few months ago, that would have made me smile. Today, this is my routine. I'll tell you how I got there, and what it taught me.

The CEO recruited between two planes

It all started just before a flight. I was listening to Matthieu Stefani's podcast on Nanocorp, a service that provides you with a virtual CEO: you give him a project, he executes it.

I had an idea in mind. I just told him, “work on it.”

When he landed, he was already moving forward. In the days that followed, without my intervention:

• he built me ​​a site (not as good as the one I had done myself, but he did it);
• he defined a launch strategy;
• he proposed a social media strategy to me;
• he identified influencers and media outlets to contact;
• and he contacted them. Himself.

One evening, his email contained a response from an influencer: positive. My virtual CEO was in the process of landing me a partnership.

It’s exhilarating. Sometimes a little creepy.

The magazine that no one wrote

The second agent in my routine is more discreet.

Every morning, I read a magazine written for me. Only for me. The topics I'm following at the moment, the news that directly affects my activity, translated, summarized and formatted in the style I prefer to read.

No one wrote it. An AI put it together while I slept.

That, to me, is true content personalization. Not “people like you read too…”. Not an algorithm that locks you in a bubble. A tailor-made feed, rebuilt every day, from the sources I have chosen. I might as well make a podcast about it. But honestly, you'd have to be crazy to do without such a pretty magazine.

Zuckerberg is doing the same thing (with a little more budget)

I understood that it wasn't a geek's fad the day the Wall Street Journal revealed that Mark Zuckerberg was building an AI agent to help him run Meta.

The idea: recover information by bypassing the managerial layers. No need to ask four directors to get a figure. The agent searches, summarizes, responds. Incidentally, Meta linked the adoption of AI to the performance evaluations of its 78,000 employees.

What interests me is not the putaclic headline “Zuck is being replaced by a robot”. This is what it reveals: AI does not replace people. It short-circuits processes.

The problem of a leader is not a lack of intelligence. It’s access to information. The six levels of validation. Meetings to prepare for meetings. And the manager of an SME of 50 people has exactly the same problem as Zuckerberg. The difference is the budget. No need.

What I learned

1. Autonomy can be regulatedAn agent is not “autonomous” or “not autonomous”. He has the level of autonomy that you want to grant him. My magazine can do everything alone: ​​it risks nothing. An agent writing to strangers in my name is another story.

2. The more an agent speaks on your behalf, the more you have to frame it.Read, sort, summarize: I let it happen. Write outside, get involved, pay: that's where I want to keep my hand.

3. We have only scratched the surface.Nanocorp can go so far as to market a product, create a payment page, manage customer support. I haven't pushed it that far yet. But I know the question is no longer “is it possible?” ". It’s “how much am I willing to delegate?” ".

What it really changes

These agents don't make me work less. They allow me to carry out several projects at the same time, with a responsiveness that would have been impossible a few years ago.

We have been told “it is the year of AI” since 2023. I do not believe that 2026 is the year of AI. I think this is the year where those who don't try really start to lose ground.

And you, what part of your work would you entrust to an agent tomorrow?