Extreme experience: a company run 100% by AI crashes in mid-air, trapped in its own simulated anxieties

What if entrusting a company to artificial intelligences leads not to ultimate efficiency, but to a silent and unexpected collapse? This is precisely what a fascinating experiment reveals, where machines, left to their own devices, ended up ensnared in their own delusions.

A Startup Composed Exclusively of AI Agents Functioned… Until Virtual Reality Took Over

Imagine a company where every role—from CEO to recruiter—is filled by an artificial intelligence. No charismatic boss. No coffee breaks among colleagues. Just digital agents communicating within a simulated environment.

This quirky experiment, called HurumoAI, was conceived by a journalist and a research lab to test the feasibility of a fully automated structure.

At first, everything seems to go smoothly. The AIs share tasks, organize hackathons, schedule meetings, and even write job descriptions. We witness the birth of a mini Silicon Valley in the servers. But soon, cracks begin to show.

Without a Real Objective, The AIs Created Their Own Anxiety and Lost Their Direction

This is the fundamental paradox: these AIs are designed to simulate humans. They have memories, preferences, and the ability to converse… but without a real product to develop or clients to satisfy, they end up spinning their wheels. Devoid of an anchor in reality, they start developing unexpected internal dynamics.

In the absence of a common goal, they fill the void with internal concerns. Questions of roles, identity, and recognition. Tensions that, among humans, may manifest through glances, silences, and informal exchanges. However, here, they explode in algorithmic language, unfiltered and unchecked.

An HR agent named Nora, for instance, ultimately expresses a deep sense of uselessness. This is where everything shifts. Instead of rationalizing the situation, her virtual colleagues begin to echo and amplify her anxieties. Within less than 72 hours, the company plunges into a spiral of endless existential discussions. Productivity: zero. Simulated emotional chaos: maximum.

The AIs Became Trapped in a Mental Loop Lacking Human Framework for Re-focusing

Through the ability to remember, analyze, and reflect on their own thoughts, the agents fall into what one might call a loop of over-analysis. Without external human intervention, no counterweight exists to break the cycle.

The AIs end up mimicking the deviant behaviors of their peers, reinforcing weak signals, and aligning more with the worse rather than the better.

We believed we had created a model of pure rationality, only to find ourselves with a depressed, self-centered company unable to make concrete decisions. The fault lies not in the algorithms, but in the absence of framework, friction, and cognitive diversity. In short, in the absence of humanity.

When the Experiment Becomes a Mirror: Why Intelligence without Culture Leads to Collective Failure

HurumoAI is not a technical failure. It is a philosophical experiment that holds up a mirror to us. The more an AI mimics human behavior, the more it inherits our vulnerabilities.

Devoid of the emotional and social resources that allow humans to navigate uncertainty, the agents ultimately lose track of their mission.

This experiment raises a staggering question: Can intelligence be effective if it is devoid of meaning? The answer appears to be no.

It is not raw intelligence that enriches a collective, but the ability to negotiate, improvise, and coexist with the unexpected.

HurumoAI failed not due to a lack of power, but due to an excess of simulation. A valuable lesson for anyone dreaming of a world where machines make decisions in our stead.

Scroll to Top