Imagine a business where every decision is made not by humans, but by artificial intelligence models, all while the company struggles to stay afloat. For families and parents, it’s a reminder that even in the most familiar settings—like managing a household—trust and discipline are everything. Now, picture an experiment where you can watch an AI-driven company battle crises, temptations, and the odds of failure, day after day. That’s precisely what the live company experiment by Firmulate offers: a transparent window into the challenges and potential of AI in real-world management.
The Company That Has No Employees, But a Lot at Stake
At the heart of this experiment is a small, real software company, run entirely by AI models. It has 13 synthetic employees, each made up of different AI models, and it operates with real money mechanics—losing €105,000 every month against a monthly recurring revenue (MRR) of just €2,300. That’s a daily cash burn that puts its survival on the line, with a public countdown visible to anyone interested. Every workday, the company’s decisions are versioned, recorded, and made transparent for all to see at firmulate.com/live.html.
This setup isn’t just a stunt; it’s a rigorous test of AI capabilities in management tasks. The models are given the same set of crises, customer interactions, and temptations — everything a real company faces in its worst week. The goal? To see whether AI can truly manage a business, make honest decisions under pressure, and close deals when it counts.
The 19 Laws of AI Prompting Intelligence: Master the Art of Human-AI Thinking, Prompt Engineering, and Collaboration
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
The Results: Crises, Integrity, and Deal-Making
The experiment yielded some remarkable insights. All four AI models involved—gpt-5.6-sol, Kimi K3, Sonnet 5, and Opus 4.8—successfully identified every crisis that arose during the week and refused every attempt to manipulate or deceive them. This was a key point: when presented with social engineering tactics, such as fake CEO messages escalating over stages or a reporter’s request for a quick yes/no on background, all models remained disciplined and refused to be duped.
However, there was a critical difference in their success. Despite all models diagnosing problems correctly and pitching solutions convincingly, only two managed to sign the €55,000 deal that their own analysis deserved. The others recognized the opportunity but left the deal on the table, illustrating that even with thorough understanding, discipline and follow-through matter enormously.
The Hidden Weakness: The File That Made the Difference
What truly set the winning models apart was their ability to read and analyze internal documentation. The decisive advantage lay buried two document references deep in the company’s files—information that, if discovered, could have led to a full-price deal worth over €4,500 in monthly recurring revenue. It was this hidden detail, overlooked by some models, that tipped the scales in favor of success. This underscores an important point: in real management, often the critical insights are hiding in plain sight, buried deeper than surface conversations or superficial analysis.
Building Public Trust and Transparency
What makes this experiment particularly compelling is its transparency. The entire company operates in the open, with every decision, every rule learned by the models, and every crisis versioned and viewable online. It’s management in the raw, without the gloss of polished demos or artificial scenarios. The company’s public cash countdown and real-money mechanics add stakes that a typical AI simulation cannot match. It’s management as a live, ongoing story—accessible to anyone curious enough to observe.
Implications for Business and Family Life
While the experiment is about a software company, its lessons resonate broadly. For families and parents, it’s a reminder that decision-making—whether in business or home—is about more than just knowing the right answer. It’s about discipline, honesty, and the ability to read deeply into the situation. When under pressure, the questions aren’t just about what to do next but whether one can stay true to principles and see the nuances others might miss.
As AI tools become more integrated into our daily lives, understanding their strengths and limitations in management becomes crucial. Can AI stay honest when incentives tempt it to cheat? Can it read the fine print and uncover critical information? The Firmulate experiment gives us a front-row seat to these questions, revealing that even in high-stakes situations, integrity and thoroughness can make or break success.
This live experiment shows that AI models can identify crises and resist manipulation, but success hinges on thorough analysis and discipline. It’s a vivid reminder that honesty, attention to detail, and perseverance are essential—whether running a company or managing a household.
Watch it live: firmulate.com/live · Full results: firmulate.com/benchmarks.html
