Imagine running a beauty salon where every decision is made by an AI, tested in real time — not in a simulation, but in a live, unfolding business crisis. This is not a sci-fi story; it’s the reality of a pioneering experiment called Firmulate. Just as you strive to perfect your craft, this company is pushing the boundaries of AI decision-making, showing what it can do — and what it cannot.
The Live Experiment: An AI-Driven Company Under Scrutiny
Firmulate is a public, transparent experiment that runs a small software company using four different advanced AI models. Each model is tasked with managing the company through its most challenging week — facing the same customers, crises, and temptations in every run. The goal? To see if AI can not only diagnose problems but also execute decisions and close deals, all while maintaining honesty and discipline.
What makes this experiment extraordinary is that it’s happening in real time, with every decision versioned, auditable, and publicly available at firmulate.com/live.html. This setup provides an unprecedented window into AI management, where every slip-up and success is openly observed.
Construction Program Management – Decision Making and Optimization Techniques
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
How the AI Performs — and What It Reveals
The results are revealing. All four AI models identified every crisis and refused manipulative tactics, such as social engineering attempts—like fake CEO messages or reporter tricks—five for five. For instance, Kimi K3 explicitly refused to sign a fake deal, citing security concerns. Despite their honesty, only two of the models succeeded in closing a €55,000 deal after their analysis. The other two diagnostics, while accurate, failed to follow through and left potential deals on the table.
This shows that AI can recognize problems and resist unethical manipulation, but executing on opportunities remains a challenge. The critical weakness was found not in customer interactions but buried within their own internal files. When models read deeper into the company’s documents—something that’s often overlooked—they uncovered a key piece of information that led to securing a profitable deal, adding over €4,500 to monthly recurring revenue.
The Human-Like Struggles of AI Management
One of the most compelling aspects is how AI models handle social engineering. Despite escalating fake CEO messages and reporter tricks over multiple stages, all models refused to take unethical shortcuts. Kimi K3, in particular, treated such requests as potential impersonation or approval-bypass, demonstrating a disciplined, security-minded approach similar to a seasoned manager.
The Company and Its Mechanics
This ongoing, publicly visible enterprise employs 13 synthetic ’employees’ — not real humans — governed by over 680 self-learned rules. It burns €105,000 per month while generating just €2,300 in monthly recurring revenue. Every workday, its decision-making process is versioned, and its cash countdown is transparent, making it a living case study for management quality in AI-driven operations.
Insights for Business Leaders
For those in beauty and personal care, the lesson is clear: when deploying AI tools, the focus should be on reliability and integrity, not just how well they generate text or engage in chat. Will your AI finish what it starts? Will it read your files thoroughly before acting? Will it stay honest under pressure? The current leaderboard, with scores of up to 95 out of 100, shows that AI can perform at a high level but still faces real-world limitations.
In this experiment, the top performers—gpt-5.6-sol and Kimi K3—demonstrated strong decision-making and integrity, closing deals at full value. Meanwhile, other models struggled with discipline, leaving opportunities unexploited even when they knew the right move.
What’s Next? Running Your Own AI Business Wargame
Business leaders can now test their own AI management strategies with a tailored, read-only version of this experiment at Firmulate’s pilot platform. It’s a risk-free way to see how your AI tools handle crises, temptations, and ethical dilemmas—before they get integrated into your critical systems.
Watch it live: firmulate.com/live · Full results: firmulate.com/benchmarks.html
