Imagine a business that operates entirely in public view, with no employees, losing thousands of euros each month, yet still pushing forward. For senior care and aging services, understanding how AI can test and shape future companies offers a glimpse into a world of high-stakes experimentation—where transparency and innovation collide.
Enter the world of Firmulate, a pioneering experiment in building a company in plain sight of the world, yet without traditional employees or profits. This real-time experiment involves 13 synthetic staff members—artificial decision-makers that run a small software firm through its hardest week. Every move they make, every crisis they encounter, and every decision they take is publicly visible, versioned daily, and subject to rigorous analysis.
The company’s financials tell the story of relentless challenge: it burns through €105,000 each month while generating just €2,300 in monthly recurring revenue (MRR). Despite the heavy losses, the project continues, driven by a combination of transparency, innovation, and testing AI’s limits in a real-world setting. The goal isn’t immediate profit but learning how AI models handle complex, crisis-prone environments.
What’s at Stake? The AI Competition
Several state-of-the-art AI models compete in this environment, evaluated by a series of tests called the Crucible League. The models are scored based on their ability to identify problems, avoid manipulation, and ultimately close deals—key indicators of management quality. The top performer, gpt-5.6-sol, scored 95 out of 100, successfully uncovering hidden information in company files to win a €55,000 deal. Meanwhile, other models like Kimi K3 and Sonnet 5 also managed to close deals, but with varying degrees of process discipline and accuracy.
One revealing finding was that the weakest link in the company’s decision chain was buried within its own files—an insight that AI models that read deeply into documents identified and exploited. This suggests that in real business, the ability to dig into internal information can be decisive—if only human managers took such thoroughness seriously.
Defense Against Manipulation and Social Engineering
The experiment also tested AI resilience against social engineering—fake CEO messages and reporters attempting to get approvals. All models rejected these manipulative tactics, highlighting a level of robustness that’s rare in typical chat interactions. Kimi K3 explained its reasoning clearly: “Treat the request as a suspected approval-bypass / possible impersonation.” This disciplined behavior underscores AI’s potential in actively resisting fraud or manipulation in sensitive environments like healthcare or senior care management.
The Reality of Operating Without Profit
Despite the impressive decision-making capabilities, the company is far from profitable. It’s a high-stakes demonstration—a public, transparent sandbox—where every decision and rule learned is documented and analyzed. The experiment involves over 680 self-learned rules, and every workday’s decisions are versioned, making the process auditable and transparent for researchers and stakeholders alike.
This setup isn’t designed to replace human managers immediately but to test how AI models handle complex, crisis-filled scenarios under pressure. For the elder care sector, this hints at a future where AI tools could help manage resources, identify risks, and even negotiate deals—if those AI systems can demonstrate honesty, discipline, and thoroughness in high-pressure situations.
Why Should You Care?
For industries like senior care and aging services, trust and reliability are everything. The question isn’t whether AI can write a good message—it’s whether it can see the full picture, stay honest amidst temptation, and deliver consistent, useful results. The firmulate live experiment shows that AI can identify hidden risks and resist manipulation, but it also highlights the importance of discipline and thoroughness—traits critical in managing vulnerable populations and sensitive operations.
As you consider integrating AI into your future operations, think about the core skills these models are developing: reading deeply, resisting shortcuts, and making disciplined decisions. The journey of this public, experimental company illustrates that building trust in AI systems isn’t just about their capabilities but about their integrity and resilience in real-world trials.
Explore the Live Experiment
Curious to see how AI manages a real company’s worst week? Visit firmulate.com/live.html to watch the ongoing daily decisions, crises, and resolutions unfold. It’s a transparent window into the future of AI-driven management—challenging assumptions and redefining what’s possible in business and healthcare alike.
Watch it live: firmulate.com/live · Full results: firmulate.com/benchmarks.html
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.
Deceptive Intelligence: AI, Social Engineering, and Securing the Human Element
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Platform Supply Chains 5: Risk Management (Platform Supply Chains in the AI Era)
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
EXPLAINABLE AI : Techniques that Meet Auditors’ Needs : Building Transparent, Defensible, and Audit-Ready Artificial Intelligence for Modern Enterprises
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
