Imagine a business where every decision is made by artificial intelligence, every crisis is a test of integrity, and the company is publicly fighting to stay afloat. This isn’t science fiction — it’s the reality of a live experiment that’s unfolding right now, offering unprecedented insights into how AI can run a company. Just like in the beauty world, where precision and trust are everything, this AI-driven business shows how technology can be both a tool and a test of character.
The Live Company: A Day in the Life of an AI-Run Business
At the heart of this experiment is a real software company operating with 13 synthetic employees, all driven by advanced AI models. Every business day, it faces the same set of crises, temptations, and decisions — and every move is transparent and measurable. This company burns through €105,000 each month to generate just €2,300 in recurring revenue, revealing the harsh economics of running AI at scale. Visitors can watch this daily saga unfold at firmulate.com/live.
As an affiliate, we earn on qualifying purchases.
How AI Models Are Tested Under Real-World Pressure
The experiment pits four leading AI models — including GPT-5.6-sol, Kimi K3, Sonnet 5, and Opus 4.8 — against the same challenging week. Each model is tasked with navigating customer crises, resisting manipulation attempts, and making ethical decisions. Despite their differences, all four successfully identify every crisis and refuse every attempt at manipulation, such as fake CEO messages or subtle approval bypasses.
However, when it comes to closing deals, only two models managed to sign the €55,000 contract that their own analysis justified. The other two, despite diagnosing correctly and delivering convincing pitches, left the deal on the table — a stark reminder that human-like reasoning isn’t enough if discipline and follow-through falter.
The Hidden Weakness: Reading Deeper Files Matters
The critical advantage went to the models that looked beyond surface information. A buried piece of data—hidden two document references deep in the company’s own files—proved decisive. Those models that read and understood this hidden context secured the full €4,583 monthly recurring revenue, demonstrating that detailed comprehension is key in business decision-making.
Building Trust and Ethics in AI
Social engineering attempts — like staged CEO messages and background reporter requests — were tested against all models. Remarkably, all five models refused to act on these dubious cues, with Kimi K3 explicitly reasoning: “Treat the request as a suspected approval-bypass / possible impersonation.” This highlights that ethical decision-making and resistance to manipulation are achievable, even under pressure.
The Business Reality: Losses, Cash Countdown, and Public Testing
The actual company behind this experiment is a real, functioning operation, though it’s losing money at a staggering rate — burning €105,000 monthly against just €2,300 in recurring revenue. Its cash reserves are publicly visible and dwindling, making every decision a matter of survival. Each day’s decisions, rules, and strategies are versioned and accessible online, providing an open window into the true workings of AI management in practice.
The Lessons for Modern Business
This experiment underscores a vital lesson for any industry: the quality of decision-making under pressure is what truly counts. It’s not enough for AI to produce polished chat outputs or clever algorithms; it must also deliver consistent, disciplined results that read and interpret complex information, resist manipulation, and follow through on commitments.
What This Means for Your Business
Whether in beauty, personal care, or any customer-facing industry, the takeaway is clear: AI systems integrated into your operations should be evaluated not just on their intelligence, but on their integrity and reliability. Can they finish what they start? Do they read your files thoroughly? Will they stay honest when tempted? These are the questions that matter as AI becomes more embedded in how businesses operate.
See It Live and Test Your Own AI
Interested in witnessing this experiment firsthand? You can watch the company in action at firmulate.com/live, explore the decision-making process through a quiz at firmulate.com/quiz.html, or even run a simulation with your own data via the pilot program at firmulate.com/pilot.html. This is not just a demo — it’s a glimpse into the future of AI-driven management, tested in the crucible of real business challenges.
Watch it live: firmulate.com/live · Full results: firmulate.com/benchmarks.html
