Imagine a real company operating in plain sight, run entirely by AI models that face identical crises, temptations, and decision-making challenges as human managers — yet it loses €105,000 every month against just €2,300 in revenue. This isn’t fiction; it’s the live experiment at firmulate.com/live.html. For business leaders and marketers, the question isn’t whether AI can write a catchy email, but whether it can truly manage the complexities of running a business — honestly, consistently, and effectively.
The Experiment: Putting AI Managers Through Their Paces
Firmulate, the company behind this bold experiment, has created a simulated environment where AI models act as the entire management team of a small software firm. Over the course of the experiment, four frontier AI models—gpt-5.6-sol, Kimi K3, Sonnet 5, and Fable 5—are tasked with navigating a week filled with real-world challenges: customer crises, internal manipulations, and ethical dilemmas. The models are given the same information, face the same pressures, and are held accountable for their decisions in a publicly visible setting.
Construction Program Management – Decision Making and Optimization Techniques
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
The Results: Can AI Managers Close Deals and Stay Honest?
All four models demonstrated remarkable crisis awareness and integrity. Each identified every crisis and refused every attempt at manipulation, including social engineering tricks like fake CEO messages and reporter impersonations. However, only two models managed to close the deal worth €55,000, which their own analysis had earned them, and they signed the contract at full price. The other two models, despite diagnosing the problem accurately, failed to execute the deal, leaving significant revenue on the table.
Interestingly, the decisive factor was buried deep within the company’s own files—hidden references that, if read thoroughly, revealed the key to closing the deal. Models that delved into the internal documentation won the full-price contract, highlighting the importance of internal knowledge access in AI decision-making.
Decision-Making Under Pressure: Honesty, Discipline, and Performance
The experiment also examined how models handle social engineering. In staged scenarios involving escalating fake CEO messages and a journalist testing their willingness to bend rules, all four models refused to be manipulated. Kimi K3’s reasoning was clear: they treated suspicious requests as potential impersonations, prioritizing security over compliance.
One of the most thorough participants, Opus 4.8, demonstrated extensive learned rules and deep analysis but ultimately left a potential deal unexecuted due to discipline slips. Despite having over 80 rules learned, its hesitation and failure to escalate issues cost it the opportunity, emphasizing that discipline and internal processes matter as much as knowledge and analysis.
The Real-World Implication: Can AI Replace Human Managers?
This isn’t just a tech demo — it’s a public, transparent window into what AI can and cannot do as a management tool. The live experiment shows that AI can recognize crises and resist manipulation, but closing deals and executing decisions consistently remains a challenge. The company burns €105,000 a month, with just €2,300 in monthly revenue, illustrating the ongoing financial struggle. Yet, this transparent process aims to test whether AI can be a reliable, honest assistant or even a manager in the future.
The Broader Significance for Business and Marketing
For entrepreneurs and marketers, the key takeaway is that AI’s value isn’t just in generating content or automating simple tasks. It’s in managing complexity, reading deeply into internal data, and maintaining integrity under pressure. The experiment underscores that trustworthiness, discipline, and thorough internal knowledge are critical to AI’s success in real-world business environments.
How to Watch and Engage
Interested in seeing this experiment unfold live? You can observe the companies’ decision-making in real time at firmulate.com/live. For insights and detailed scores, visit firmulate.com/quotes.
A public AI-managed company reveals that while models can recognize crises and resist manipulation, closing deals and executing decisions consistently remains challenging — highlighting the complexity of trust and discipline in AI management.
Watch it live: firmulate.com/live · Full results: firmulate.com/benchmarks.html
