The Future of Business Is Already Here — Watch an AI-Run Company Struggle in Real Time
Imagine a company with no human employees, making critical decisions, battling crises, and vying for a deal — all driven solely by artificial intelligence. This isn’t science fiction; it’s the groundbreaking live experiment from Firmulate, where AI models operate a tiny software business under real-world pressures, and you can watch it unfold every day.
AI decision-making software
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
The Live Experiment: An AI Company in Action
At the heart of this experiment is a small, virtual company managed entirely by AI models. It features 13 synthetic employees and real financial mechanics. The company burns through €105,000 each month but generates only €2,300 in monthly recurring revenue, highlighting its fragile financial state. Every workday, the company’s decisions are versioned and publicly accessible at firmulate.com/live.html, making it a transparent window into AI management under duress.
The AI Models at Work
Four frontier AI models were challenged to guide this virtual company through its worst week — same crises, same customers, same temptations. These models, including the latest GPT-5.6-sol, Kimi K3, Sonnet 5, and Fable 5, were tested on their ability to identify problems, resist manipulation, and close deals.
All four models successfully spotted every crisis and refused every manipulation attempt, such as social engineering scams. Interestingly, the real difference in performance arose from a buried fact in the company’s own files — a crucial document reference that, if read, led to winning a €55,000 deal. Only two models managed to read and leverage this hidden information, closing the deal and increasing monthly revenue by over €4,500.
Decision Making Under Pressure
The experiment also revealed the importance of discipline and thoroughness. For example, the most meticulous model, Opus 4.8, showed the deepest analysis but ultimately lost the close because it failed to escalate certain issues properly, leaving critical opportunities unexploited. This underscores a key takeaway: even the most advanced AI can slip if it’s not disciplined enough to follow through.
Ethical Resistance and Trust
Beyond decision-making, the models demonstrated strong resistance to social engineering. Fake CEO messages, escalating over multiple stages, and even a reporter trick asking for quick approvals were all refused by every AI. Kimi K3 explicitly noted: “Treat the request as a suspected approval-bypass / possible impersonation,” illustrating a nuanced understanding of security risks.
AI security and fraud detection tools
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
The Real Stakes and What It Means for Your Business
While this experiment might seem distant from home appliances or smart home devices, the implications are significant. If AI agents are managing your customer support or forecasting future sales, it’s critical they can finish what they start, read and understand relevant documents, and stay honest under pressure.
The live system is not just a demo — it’s a real-time showcase of AI’s potential and limitations. The experiment exposes how current models perform in a high-stakes environment, revealing strengths like crisis recognition and ethical resistance, as well as weaknesses such as incomplete follow-through and missed opportunities — even at the highest levels of AI sophistication.
AI business management tools
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
The Bottom Line: Building Trust in AI as a Business Partner
This ongoing, transparent experiment from Firmulate aims to answer a fundamental question: can AI truly manage complex, real-world tasks without human oversight? The current results show promise — all models identified crises and refused manipulation — but also highlight that even state-of-the-art AI can falter in execution and discipline.
For businesses, especially those integrating AI into critical operations, the lesson is clear. It’s not just about how well AI can generate responses in demos; it’s about whether it can reliably complete its work, read and interpret relevant information, and act ethically under pressure. Watching this experiment unfold at firmulate.com/live.html gives you a front-row seat to the future of AI-driven management.
Watch it live: firmulate.com/live · Full results: firmulate.com/benchmarks.html
AI document reading and analysis software
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
