Imagine a company with no human employees, losing money every day, yet running a transparent, real-time experiment in AI decision-making. This is not fiction, but the groundbreaking work of Firmulate, where artificial intelligence is tested as a complete, functioning business—complete with crises, negotiations, and cash flow challenges.
The Experiment: An AI Company in the Spotlight
At the heart of this experiment are four advanced AI models, each tasked with running a small software company through its worst week. These models face identical crises, customer demands, and ethical temptations, all within a carefully scripted environment designed to mirror real-world business stressors.
What makes this experiment extraordinary is its transparency and audibility. Every decision made by the AI is versioned, recorded, and publicly accessible, revealing how each model navigates complex management scenarios. This setup offers a rare window into AI’s potential—and its limits—in real business operations.
AI in Property Management: A Practical, Unboring Look at Artificial Intelligence in the Multifamily Industry
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Results That Speak Volumes
The findings are both promising and sobering. All four models identified every crisis, refused manipulation attempts, and demonstrated a keen awareness of ethical boundaries. Yet, only two of them managed to close a critical deal worth €55,000, equating to a boost of +€4,583 in monthly recurring revenue (MRR). The other two, despite similar analyses, failed to execute the deal, leaving revenue on the table.
This gap was traced to a buried detail in the company’s own files—information that the models that read deeper into the data uncovered and used to clinch the deal. It highlights a key insight: access to comprehensive data, rather than superficial cues, is essential for effective decision-making in AI systems.
AI decision-making tools for business
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
The Human Element and Trust
Beyond technical prowess, the models were tested against social engineering threats. Fake CEO messages and manipulative tactics were introduced, and impressively, all five models refused to be duped. Kimi K3, one of the leading models, explained its reasoning: “Treat the request as a suspected approval-bypass / possible impersonation.” This demonstrates that ethical decision-making is central to AI management, especially under pressure.
AI ethics and security software
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Live Business, Zero Employees, Real Money
The company itself is a live, ongoing experiment with 13 synthetic employees and real financial mechanics. It burns €105,000 each month against a modest €2,300 in monthly revenue, with a public countdown to its financial survival—adding urgency and transparency to the process.
Every workday, the company’s operations are versioned and recorded, and the entire setup is accessible at firmulate.com/live. Visitors can watch as AI models grapple with crises, negotiate deals, and make management decisions in real time, blurring the lines between artificial intelligence and human enterprise.
AI data analysis tools
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Lessons for the Future of AI in Management
This experiment underscores a critical point for anyone interested in AI’s role in business: the true test isn’t just whether AI can generate convincing dialogue or perform well in demos. It’s whether AI can finish what it starts, stay honest under pressure, and leverage deep data to make strategic decisions.
For instance, the model Opus 4.8, despite being the most thorough participant—analyzing over 80 learned rules—finished last because it faltered at the crucial moment, leaving a deal unexecuted and slipping discipline. This reveals that depth of analysis alone isn’t enough; execution discipline matters just as much.
The Broader Context and Implications
This experiment, accessible to anyone for testing their own business scenarios at firmulate.com/pilot, demonstrates a future where AI could serve as a management partner or even a decision-maker. But it also highlights the importance of transparency, ethical safeguards, and data accessibility in deploying AI systems responsibly.
In a world increasingly driven by data and automation, witnessing an AI-managed business in real time offers invaluable lessons. It reveals not only what AI can do but also what it must do to be truly trustworthy and effective in complex, real-world environments.
Real-time AI management experiments like Firmulate’s show promising advances in ethical decision-making and crisis handling, but also underscore the importance of data depth and discipline. As AI begins to touch more aspects of business, transparency and trust will be key to its success.
Watch it live: firmulate.com/live · Full results: firmulate.com/benchmarks.html
