Imagine a business with no human employees, running in real-time, losing money every single day, yet still actively competing in a simulated marketplace. For sports fans and strategists alike, this ongoing experiment offers a rare glimpse into the future of AI-driven management and decision-making. Welcome to the live world of Firmulate, where AI models operate a virtual company under relentless pressure, and every decision is open for public scrutiny.
The Real-Time Business War Game
At the heart of this experiment are four advanced AI models, each tasked with running a small, software-focused company through its most challenging week. These models—GPT-5.6, Kimi K3, Sonnet 5, and Fable 5—are not just chatbots; they are decision-makers in a live environment, facing the same crises, customer demands, and ethical temptations as real managers.
What makes this experiment unique is its transparency: every decision, every crisis response, and every attempt at manipulation is versioned and auditable. The company’s financials tell a stark story—burning €105,000 each month against a mere €2,300 in monthly recurring revenue. With a public cash countdown visible at firmulate.com/live.html, viewers can watch the company’s struggle unfold in real time, every workday being recorded and versioned.
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.
Decisive Findings: AI’s Strengths and Limits
The models demonstrated impressive vigilance: all four spotted every crisis presented to them. Whether it was customer pressure, ethical dilemmas, or manipulation attempts, they refused to bend or cheat. For instance, when faced with an escalating fake CEO message and a reporter trick, all five models refused to cooperate, citing concerns over impersonation and bypassing approval protocols—an indication of their built-in safeguards.
However, the experiment also revealed critical weaknesses. Only two of the four models managed to close the deal that their analysis had identified as the right move—signing a €55,000 contract that would have added €4,583 in monthly revenue. The other two, despite making the correct diagnosis, either left the deal on the table or failed to act decisively, illustrating that understanding and execution are not always aligned for AI managers.
AI Entrepreneur’s Handbook: Build a Profitable Business and Make Money by Unleashing the Power of ChatGPT and Artificial Intelligence (Includes 150+ ChatGPT prompts to turbocharge your business)
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
The Hidden Vulnerability: The Buried Fact
Remarkably, the decisive edge came from reading a document buried two references deep in the company’s internal files—information that was not immediately apparent from external customer interactions. The models that delved into these internal files successfully identified a crucial fact, leading to the full-price deal. This points to an often-overlooked aspect of AI decision-making: the importance of deep internal knowledge, not just surface-level customer interactions.
The AI-Native Technology Leadership Playbook: A Practical Guide to Building AI-Native Platforms, Enterprise Products, and High-Performance Technology Teams
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Implications for Business and Sports Strategy
For sports fans and strategists, this experiment offers a parallel: just as a team’s success hinges on internal data, tactics, and ethical play, so too does this AI-driven company. The models’ ability to avoid manipulation and discover hidden information echoes the importance of internal game tape and strategic insights in sports management. It suggests that AI can be a disciplined, honest actor capable of strategic depth—if given the right data and structure.
50 ML Projects To Understand LLMs: Investigate transformer mechanisms through data analysis, visualization, and experimentation
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
What This Means for the Future
While this experiment is a small, controlled simulation, its implications are vast. As AI begins to touch more aspects of business—CRM, customer support, forecasting—the questions extend beyond chat quality to fundamental issues of trust, discipline, and execution. Can AI manage under pressure? Will it stay honest when tempted? And crucially, can it deliver useful work that justifies its cost?
Currently, the live experiment demonstrates that AI can identify crises and refuse manipulative tactics. Yet, the failure to close some deals despite correct diagnoses highlights that management and execution still have room to grow. For stakeholders and strategists, the takeaway is clear: AI management is not just about intelligence, but about discipline, internal knowledge, and integrity.
The Call to Action
Businesses and sports organizations alike should consider the value of “wargaming” their AI agents before fully deploying them. Using tools like this live platform, leaders can simulate crises, test decision-making, and identify weaknesses before real-world consequences unfold. This ongoing venture by Firmulate is a step toward transparent, disciplined AI management—crucial as AI begins to take on more operational roles in our lives.
This live experiment reveals that AI can spot crises and refuse manipulation, but managing and executing deals remains a challenge. Transparency and internal data access are key to trustworthy AI management—vital lessons for business and sports strategy alike.
Watch it live: firmulate.com/live · Full results: firmulate.com/benchmarks.html
