Imagine a company with no human employees, burning through €105,000 every month yet only earning €2,300 in recurring revenue. It’s a real, live experiment, unfolding publicly at firmulate.com/live. This isn’t fiction — it’s a fascinating peek into how AI models manage a company under extreme conditions, revealing what it really takes for artificial intelligence to operate in the complex world of business.
The Company in the Public Eye
The experiment features a fully functioning, synthetic workforce—13 AI ’employees’—each guided by over 680 self-learned rules that govern decision-making, strategy, and crisis management. Every workday, the company’s operations are versioned and made transparent for public inspection. Its core mechanics mimic real-world business challenges: customer crises, ethical dilemmas, and the temptation to manipulate processes.
AI decision-making software for business
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Testing AI Under Fire
The central experiment pits four advanced AI models against the same week of business chaos. Each model faces identical scenarios: customer complaints, internal crises, and external pressure. The goal? See how they perform, whether they can identify hidden opportunities, and if they can maintain integrity under stress.
All four models successfully spotted every crisis and refused every manipulation attempt—including social engineering tricks like fake CEO messages and reporters’ inquiries. This suggests a high level of discipline and robustness across the board.
AI cybersecurity tools for companies
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Decisive Factors and Hidden Weaknesses
Despite their vigilance, only two models managed to close the deal worth €55,000, thanks to a buried detail in internal files that was overlooked by others. The models that read and interpret internal documents at a deeper level were able to uncover missed opportunities, resulting in a significant boost to their recurring revenue potential (+€4,583 monthly).
AI document analysis software
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Real Money and Real Risks
The company’s financials are stark: it spends €105,000 per month but earns just €2,300 in recurring revenue. The live site intensively tracks this cash countdown, adding a layer of urgency and realism that pure chat demos cannot replicate. Every decision made by these AI ’employees’ is logged, versioned, and subject to public scrutiny, emphasizing transparency and accountability in AI management.
AI crisis management platform
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Behavioral Traits and Discipline
The most thorough participant, the Opus 4.8 model, analyzed over 80 learned rules and conducted deep assessments, yet it finished last. It left critical deals unexecuted and slipped into internal inefficiencies, demonstrating that deep analysis alone doesn’t guarantee success. Interestingly, the other models, running at a high effort parameter, performed better in closing deals, suggesting that discipline and adherence to process are critical.
Implications for Business and AI Adoption
This experiment highlights crucial questions for companies considering AI integration: can these models finish what they start? Do they read and interpret internal documents to find hidden opportunities? Can they resist manipulation and maintain integrity? In the context of smart home and appliances, where AI manages not just data but physical systems, these questions are equally vital.
Beyond the Demo: Wargaming Your AI Workforce
Business leaders can run their own ‘wargames’ against a read-only export of their operations, testing AI decision-making under simulated crises. This approach, showcased at firmulate.com/pilot.html, allows companies to evaluate AI’s performance without risking actual systems or data.
Why This Matters for Your Smart Home
While this experiment may seem distant from smart home tech, the underlying lessons are pertinent. Just as AI models in this experiment must stay honest, finish tasks, and read critical information, smart home systems depend on reliable, ethical AI to manage security, energy use, and safety. Understanding AI’s true capabilities and limitations is vital as these systems become more integrated into our daily lives.
This live experiment exposes the strengths and weaknesses of AI decision-making in a high-stakes environment. For consumers and businesses alike, the key takeaway is: the future of AI isn’t just about smarter algorithms—it’s about trustworthy, disciplined systems that can truly deliver on their promises, especially when it counts most.
Watch it live: firmulate.com/live · Full results: firmulate.com/benchmarks.html
