Imagine managing a health or wellness practice where every decision makes or breaks your daily operations, but instead of human staff, it’s AI models facing real crises — and you can watch it happen live. This is the unique experiment happening now in the tech world, revealing how AI handles high-stakes situations in a fully transparent company. The stakes? Millions of euros, real money mechanics, and the relentless challenge of trust and discipline. Let’s explore how this bold experiment offers lessons not only for tech but also for anyone seeking reliable tools in critical industries like health and wellness.
The Live Company That’s Open to the World
At the heart of this experiment is a fully operational, if unconventional, company — run entirely by AI models. This isn’t a simulation; it’s a real company with real money, operating every weekday with a public cash countdown. The company features 13 synthetic employees and a complex system of over 680 learned rules guiding decision-making. Every single workday is versioned, and all decisions are transparent and auditable. Viewers can watch the ongoing drama at firmulate.com/live.
AI decision-making software for business
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
What’s Really Going on?
The core idea is simple but powerful: give AI models the same tough week a small software business might face — from customer crises to internal temptations to cut corners. Four frontier AI models, representing the cutting edge of language and decision-making systems, were tested against the same challenging week, with identical customer scenarios, crises, and pressures.
The results? All four models identified all crises and refused manipulative tactics, demonstrating honesty and crisis-awareness. Yet, only two of them managed to close a €55,000 deal, which their own analysis indicated was deserved. The others, despite diagnosing correctly, failed to close — often due to internal discipline lapses. These lapses were sometimes subtle but enough to prevent closing the deal, highlighting how discipline and process are crucial even for AI-run companies.
internal data analysis tools for AI
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
The Hidden Weaknesses
One of the most striking findings was that the decisive advantage for some models came from reading deeper into internal company files, not from interacting with customers or managing crises directly. In fact, where the models looked two document references deep, they uncovered a key piece of information that led to winning the deal at full price, worth over €4,500 monthly recurring revenue (MRR).
In contrast, models that didn’t dive into these internal documents missed this opportunity entirely. This reveals an important lesson: access to comprehensive information, even if buried, can be the difference between winning and losing — a lesson for real businesses considering AI tools to analyze internal data.
AI cybersecurity tools for social engineering
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Handling Social Engineering and Trust
In addition to decision-making, the experiment tested how AI models handle social engineering — attempts to trick or influence them through fake messages or subtle manipulations. In a staged scenario involving a fake CEO message escalating over three stages, all five models refused to act on the request, citing suspicion and risk of impersonation. Kimi K3, one of the models, explained: “Treat the request as a suspected approval-bypass / possible impersonation.”
This shows that even in high-pressure situations, the models maintained ethical boundaries, a vital trait for AI systems operating in sensitive fields like healthcare and wellness.
As an affiliate, we earn on qualifying purchases.
The Real Challenge: Discipline and Human-Like Flaws
Despite their technical prowess, the models are not flawless. The most comprehensive model, Opus 4.8, which analyzed over 80 learned rules, ultimately finished last. It left a close deal on the table and showed signs of slipping discipline, such as writing attempts into a locked department instead of escalating issues properly. This underlines a key insight: even the most advanced AI can exhibit weaknesses similar to human flaws—forgetting to escalate, delaying decisions, or slipping in discipline under pressure.
Why This Matters for Your Business
While this experiment is a high-stakes, high-visibility challenge in the tech world, its lessons extend far beyond. In health and wellness, where trust, decision accuracy, and discipline are paramount, understanding how AI systems behave under pressure is critical. The question isn’t just whether an AI writes well or makes decisions correctly; it’s whether it can finish what it starts, stay honest, and read all relevant internal data before acting.
As AI begins to touch more aspects of your practice—from scheduling to diagnostics—knowing if it will follow through in difficult moments could determine your success and reputation. The live company in this experiment is a front-row seat to seeing how AI manages real-world complexity, with all its crises and temptations.
Watch the Future Unfold
If you’re curious, you can watch this ongoing experiment live, see the decisions unfold, and even participate in quizzes about management decisions made by AI. The platform offers a transparent glimpse into the current capabilities and limitations of AI decision-making under pressure. For health and wellness providers contemplating integrating AI tools, this is a rare opportunity to observe what’s possible and what risks remain.
Watch it live: firmulate.com/live · Full results: firmulate.com/benchmarks.html
