Watch a Virtual Business Fight for Survival — Powered by AI, Losing €105K Monthly, and Still Publicly Running

Imagine running a business that’s losing more than €100,000 a month, yet you’re still operating in full view, making decisions in real time. For cleaning and maintenance companies, this might seem distant — but it’s precisely what a groundbreaking experiment is revealing about the potential and peril of AI-powered management.

The Live Experiment: A Business Under Extreme Observation

At the heart of this story is a live, real-world business simulation created by Firmulate. It’s not a game or a demonstration — it’s a functioning, everyday company with a twist. Instead of human employees, it employs 13 synthetic ’employees’ driven by advanced AI models. The company burns through €105,000 each month but generates only €2,300 in monthly recurring revenue, leaving it in a dire financial position. Despite this, it continues to operate publicly, with every workday versioned and transparent.

Watchers can see the company’s decisions unfold at firmulate.com/live. This ongoing experiment puts AI models through their paces against the real-world crises and temptations that any business faces, but with unprecedented transparency and accountability.

The AI Models and Their Performance

The experiment tests four different frontier AI models, each with a unique profile. The models are evaluated based on their ability to handle crises, avoid manipulation, and close deals — all while operating with the same data and challenges. These models read and analyze the company’s files, diagnose issues, and attempt to secure business deals.

According to the latest results, all four models identified every crisis and refused any manipulative or social engineering tactics. Notably, only two of these models managed to close a key deal worth €55,000, matching the company’s own analysis and diagnosis. The other two, despite similar decision-making and pitches, did not secure the contract. This reveals a crucial insight: the difference in performance was rooted in the models’ ability to read and interpret the company’s internal documents, which held the real opportunity — a buried fact crucial to closing the deal.

What Does This Mean for Businesses?

This experiment is more than a technical showcase; it offers vital lessons for any enterprise, including cleaning and facility maintenance firms contemplating AI integration:

• Decision Quality Under Pressure: All models identified crises, but only some made the right decisions to capitalize on opportunities.
• Reading Critical Data: The success hinged on reading deeper into internal files. For service companies, this underscores the importance of data access and analysis — knowing your own workflows and hidden insights can be the difference between winning or losing business.
• Honesty Matters: When social engineering tactics were introduced, all models refused to be manipulated. Integrity in decision-making remains vital, especially where financial stakes are high.
• Transparency and Versioning: Every decision is versioned and auditable, creating a clear trail of actions. This level of transparency could revolutionize compliance and accountability in service operations.

Real Money, Real Consequences

The live business is losing €105,000 every month yet continues operating, with a public cash countdown that viewers can watch unfold. Every decision, every crisis, every deal — or failure to close — is happening in real time, providing a rare window into how AI-managed companies perform under constant financial strain.

Notably, the company’s decision-making is guided by a set of over 680 self-learned rules — a playbook that evolves daily. Each workday is versioned, ensuring ongoing transparency. This rigorous approach allows observers to understand how AI models behave in a real-world, high-pressure environment, far beyond scripted demos or controlled tests.

The Implications for Your Business

For companies in cleaning or maintenance, the lessons are clear:

• Real decision-making involves reading and analyzing your own data to uncover hidden opportunities.
• Honest AI systems refuse manipulation attempts — a crucial trait in maintaining integrity.
• Transparency in decision processes enhances trust and accountability, especially when stakes are high.
• Understanding what makes an AI’s decision effective can save you from costly mistakes and missed deals.

While the experiment might seem distant, its core questions are universal: Will your AI read deeper into your business files? Will it stay honest under pressure? Can it finish what it starts? As this live company demonstrates, these are the questions that will determine the future of automation in service industries.

See It Live and Decide for Yourself

Curious to see this kind of decision-making in action? The experiment is ongoing, and you can watch the company run every business day at firmulate.com/live. Try the quiz to test your understanding of AI decision-making at firmulate.com/quiz, or explore how your own enterprise could run the same wargame with a read-only export at firmulate.com/pilot.html.

Watch it live: firmulate.com/live · Full results: firmulate.com/benchmarks.html

Powered by Thorsten Meyer AI