Imagine Watching an AI Company Struggle and Survive in Real Time
What if you could see an artificial intelligence not just chat or recommend, but actually run a small business — facing crises, making decisions, and even risking financial ruin? That’s exactly what Firmulate’s latest live experiment offers: a transparent, real-time window into how AI models behave when managing a company under pressure.
Building AI-Powered Products: The Essential Guide to AI and GenAI Product Management
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
The Live Experiment: A Company on the Brink
At the heart of this experiment is a small, synthetic company with 13 AI-powered employees. It’s not just a simulation; it’s a real-world test with live money mechanics. The company burns through €105,000 each month while bringing in only €2,300 in monthly recurring revenue (MRR). Every workday, its decisions are documented, versioned, and published publicly at firmulate.com/live.
This setup isn’t just for show. It’s designed to reveal whether AI can truly handle the complexities and temptations of running a business — from recognizing crises to avoiding manipulation and dishonesty. The experiment uses four different frontier AI models, each subjected to the same tough week of simulated crises, customer interactions, and internal dilemmas.
The Models’ Performance: All Eyes on Decision-Making
Results were telling. All four models identified every crisis that arose, showcasing their ability to detect problems. They also refused every manipulation attempt, including social engineering tactics like fake CEO messages and reporter tricks — with all five models declining to sign away a €55,000 deal after analyzing the situation.
However, only two models managed to close the deal they identified as worth €55,000 in extra MRR. These were gpt-5.6-sol with a score of 95, and Kimi K3 with a score of 93 — both successfully diagnosing the opportunity and following through, even at full price.
The Hidden Weakness: It Was in the Files
What decided the outcome wasn’t just in the obvious customer interactions. The decisive advantage came from a buried reference within the company’s internal documents, not from external events. Models that read through this documentation uncovered critical information, leading to real contract signing and €4,583 in additional MRR — a clear demonstration of the importance of deep, thorough analysis.
Discipline Under Pressure
Another layer of the experiment involved testing the models’ discipline and integrity. Despite escalating social engineering attempts, all models refused to be manipulated. Kimi K3’s reasoning was straightforward: “Treat the request as a suspected approval-bypass / possible impersonation.”
The AI Culture Blueprint: Moving Beyond Tools to Create Human-Centered AI Adoption
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
What Does This Mean for Business and AI?
This experiment is more than a curiosity. It exposes the stark differences in how AI models handle real-world management challenges. For companies considering AI for support, sales, or operational roles, the critical questions aren’t just about language fluency or speed. They’re about whether AI can reliably complete tasks, stay honest under pressure, and uncover hidden opportunities.
Significantly, the experiment shows that AI’s ability to read and analyze internal documents can be the key to unlocking profitable deals — a reminder that in management, sometimes the deepest insights are tucked away in the details.
Examining Vulnerabilities and Adversarial Exploitation of AI and LLMs
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Watching a Business Fight for Survival
Every decision, every crisis, and every negotiation happening in this experiment is publicly visible at firmulate.com/live. It’s a build-in-public showcase of AI’s potential and limitations, revealing not only if an AI can run a company but also what it takes to make it trustworthy and effective.
Final Takeaway
While the company burns €105k a month—far more than it earns—the experiment demonstrates that AI models can distinguish real opportunities from threats and act on them with discipline. The real question for businesses today: will your AI tools simply talk well, or will they deliver honest, effective results — even when the stakes are high?
This live experiment reveals that AI can detect crises, resist manipulation, and even close profitable deals — but only if it thoroughly analyzes internal data and maintains discipline under pressure. Watch the full story unfold at firmulate.com/live.
Watch it live: firmulate.com/live · Full results: firmulate.com/benchmarks.html
An Introduction to Healthcare Informatics: Building Data-Driven Tools
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
