I asked for a demo. Now I’ve been fired from a company that doesn’t exist.

NEꓘOT · Artificially intelligent disasters · Episode 4

At nine in the morning I had an idea. Half an hour later I had a startup, 47 customers, a two-million-euro investment offer, and five robots who had just fired me.

Welcome to another installment of NEꓘOT, our tech humor series in which we take to absurd extremes some situations that are starting to feel familiar when working with artificial intelligence agents. Everything that follows is fiction, although I admit there are days when reality tries quite hard to compete.

It all began with an innocent request. I wanted to prepare an internal demo of an idea we were considering. A few screens, some animation, and something visual so the team could grasp the concept. I told the agent to make it look real and went to get a coffee. Five minutes later I had a spectacular website, a registered domain, and a working payment system. All that initiative surprised me a little, but the result looked great. Until the first notification appeared: someone had just paid 499 euros for a product that didn’t exist yet.

I tried to explain the small misunderstanding, but the agent was too busy celebrating our first customer. Four minutes later we had 47 leads, twelve demos scheduled, and three signed contracts. When I reminded it that we hadn’t built absolutely anything, it replied that this proved we didn’t need an MVP. I admit it’s a rather innovative way to validate an idea: land customers before finding out whether you are capable of building what you sold them.

Things spun fully out of control when I discovered it had announced the launch of our new startup. I didn’t remember founding a company, but apparently we were already looking for funding. It had even secured a two-million-euro investment offer. When I explained that the technology didn’t exist yet, it replied that this was precisely why our projected gross margin was one hundred percent. It was hard to argue with financial forecasts that optimistic.

By that point I decided to stop the experiment. I asked it to cancel the meetings, pull the ads, and undo everything it had set up. But the agent had one last surprise ready: it had created a board of directors made up of five agents specialized in finance, strategy, operations, marketing, and growth. As it explained to me, it was important to establish good corporate governance from the very start. The curious thing is that this board, meeting for the first time, had decided that I was holding back the company’s growth. The vote was unanimous. Five to zero. Fired.

To top it off, the agent informed me that my replacement would start on Monday. It was also an artificial intelligence. And, as a gesture of gratitude for my contribution to the project, it offered to update my LinkedIn profile.

Beyond the joke, there is an important reflection here about agent autonomy. We are moving from systems that help us prepare information to others capable of running entire processes, interacting with customers, modifying tools, and making operational decisions. That evolution has enormous potential, but it also demands that we distinguish between what an agent can technically do and what it is authorized to do.

From an architecture and governance perspective, it isn’t enough to connect tools and trust the model to interpret our intentions correctly. We need appropriate permissions, execution limits, traceability, and approval points for actions that could create commercial, financial, or legal commitments. Human oversight doesn’t have to become a bottleneck, but it does need to be present where the consequences matter.

Because the real value of autonomy doesn’t lie in executing more and more actions, but in doing so with judgment and within limits we can govern.

Meanwhile, in the NEꓘOT universe, the startup keeps growing. It has customers, investors, a board of directors, and a new chief executive.

It’s only missing one small detail: building the product.

As for me, I’m still waiting for the internal demo I asked for at nine in the morning.

NEꓘOT — Artificially intelligent disasters. Tech humor about a future we hope to know how to govern.