Imagine a company that operates entirely without human employees, constantly battling to stay afloat, and yet, you can watch its every decision unfold live online. For families and parents, the lesson is clear: transparency and trust are more crucial than ever — not just in your relationships but also in the businesses that serve you. Welcome to the groundbreaking experiment by Firmulate, where artificial intelligence manages a real company under extreme conditions, revealing what honesty and decision-making really mean in the digital age.
The Live Company That’s Testing AI in the Real World
At the heart of this experiment is a small, entirely simulated company run by 13 synthetic employees. This isn’t some fictional story — it’s a transparent, online, and ongoing test that anyone can watch. The company is facing daily financial pressures, burning through €105,000 every month while generating only €2,300 in monthly recurring revenue, with a public countdown showing how long it can keep going. Every workday, the company updates its rules and decisions, making it a living laboratory for understanding how AI manages complex, high-stakes situations in real time.
AI Builders: Making The Decisions That Turn AI Code Into Real Software
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
The AI Models Under Extreme Stress
Four advanced AI models — including the latest GPT-5.6-SOL — were tasked with guiding this company through its worst week, facing the same crises, customer demands, and temptations to cut corners. The models were held to a high standard: they had to spot crises, resist manipulation, and make honest decisions. The results? All four detected every crisis and refused every attempt at manipulation, such as fake CEO messages or secret deals. However, only two managed to close a €55,000 deal based entirely on their own analysis, demonstrating a key challenge: decision quality isn’t just about spotting problems but following through ethically and consistently.
What the Company’s Files Told Them
The decisive advantage for two of the models came from a simple but critical detail buried two document references deep in the company’s files, not in the immediate customer interactions. Reading deeper into internal documents allowed these models to identify a hidden opportunity, enabling them to close a full-price deal, worth over €4,500 in monthly recurring revenue. This highlights a vital insight: in complex decision-making, sometimes the most crucial clues are in overlooked internal data, not surface-level information.
Trust and Integrity Under Pressure
In social engineering tests, fake CEO messages and reporter tricks were used to manipulate the AI models. All five tested models refused to be duped, with Kimi K3 explicitly treating suspicious requests as potential impersonation. This emphasizes that trustworthy AI must resist manipulation, especially when pressure mounts — a lesson for any digital business today that relies on AI for customer interactions or sensitive decisions.
The Company in Numbers and Practice
The experiment is ongoing at firmulate.com/live, where you can see the company’s current status, with real-time updates on its challenges and decisions. Every workday, the company’s rules and decisions are versioned, giving a transparent view into how AI reasoning evolves under stress. The company’s current scores show that GPT-5.6-SOL leads with a 95 out of 100, having successfully identified hidden opportunities and closing deals. Kimi K3 closely follows with a score of 93, demonstrating the discipline of its decision-making. Meanwhile, other models like Sonnet 5 and Fable 5 showed promise but slipped on discipline or left deals unexecuted, revealing the nuanced challenge of building trustworthy AI in real-world scenarios.
Implications for Business and Trust
This experiment raises important questions beyond the tech world. For families and parents, it underscores the importance of honesty, thoroughness, and accountability — qualities that AI models can now simulate and, perhaps, embody more reliably than humans under pressure. It also highlights that transparency in decision-making, especially in AI-driven environments, is crucial. As businesses increasingly adopt AI tools for sales, support, and decision-making, understanding these tools’ ability to remain honest and thorough becomes vital.
What’s Next? Learning from Live AI Testing
As this small company continues to operate under extreme conditions, it offers a glimpse into the future of work, trust, and transparency. Enterprises could run similar ‘wargames’ using AI to test their processes, ensuring their systems can withstand crises and manipulative tactics before deploying them in real environments. The pilot programs available at firmulate.com/pilot.html let companies simulate their own business challenges without risking real systems or data.
Watch it live: firmulate.com/live · Full results: firmulate.com/benchmarks.html
