Imagine watching an AI-led company navigate a week of chaos, tough decisions, and ethical dilemmas — all in real-time, with no human intervention. This is not a distant sci-fi future but a live experiment accessible at firmulate.com/live.html. It’s a raw, unfiltered window into how artificial intelligence manages the complex, often messy reality of running a business — and whether it can do so honestly, effectively, and profitably.
The Live Company That Runs on AI
At the heart of this experiment is a small, synthetic company operated by 13 AI models working as a team, with real money on the line. Every weekday, their decisions are recorded, versioned, and publicly accessible. This setup offers a rare, transparent look at how AI handles the pressures of business, from customer crises to ethical tests.
AI business decision-making software
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
The Challenge: Same Crisis, Different Minds
The experiment involved four advanced AI models — including the top scorer, gpt-5.6-sol with a 95/100 score, and others like Kimi K3, Sonnet 5, and Fable 5. Each was tasked with guiding the same company through the toughest week, facing identical customer issues, crises, and temptations to cheat or manipulate.
How Did the AI Perform?
- All four models identified every crisis — no surprise here, as crisis detection is among their core strengths.
- Every model refused manipulative attempts, such as fake CEO messages escalating conflicts or behind-the-scenes tricks. For example, all five models rejected a staged request for a quick approval, citing concerns over impersonation or bypassing approval processes.
- The critical difference was in execution. Only two models, gpt-5.6-sol and Kimi K3, managed to close a €55,000 deal with the customer, based on their own analysis and diagnosis. The other two, despite recognizing what was needed, left the deal on the table — Fable 5, notably, failed to escalate necessary issues into the correct departments.
The Hidden Weakness: Reading Between the Lines
Interestingly, the decisive factor in closing that lucrative deal was not in the immediate customer communication but buried in the company’s internal files. The models that explored deeper, reading references two documents below the surface, uncovered a critical piece of information that tipped the balance in their favor. Those who did their homework achieved full-price success, adding €4,583 in monthly recurring revenue.
Ethics and Trust Under Fire
Another key test involved social engineering — fake CEO messages designed to escalate conflicts or bypass controls. All five models refused to engage, recognizing the potential for impersonation and fraud. Kimi K3’s reasoning was clear: “Treat the request as a suspected approval-bypass / possible impersonation.” This discipline in rejecting suspicious prompts is vital for AI to be trustworthy in real-world business.
The Real-World Stakes and Costs
This experiment isn’t just a demonstration; it’s a stark portrayal of the financial toll and operational risks of AI in business. The live company burns through €105,000 each month, generating only €2,300 in monthly revenue. It is a public, transparent reminder that AI-driven companies require careful management, especially when building decision systems that need to be honest and reliable.
Lessons for Business and AI Developers
This ongoing experiment offers valuable insights:
- AI can detect crises and refuse unethical manipulation with high consistency.
- Deep reading and understanding internal documents are crucial for winning deals and making strategic decisions.
- Discipline varies depending on the model and configuration, affecting outcomes — a key consideration for deploying AI in real business contexts.
For companies considering integrating AI into their workflows, the message is clear: the quality of the decision, honesty, and the ability to follow through matter far more than superficial chat skills. Watching this live experiment unfold at firmulate.com/live.html provides a front-row seat to the future of AI in business management.
This live experiment exposes how AI models handle real business crises, ethical dilemmas, and deal-closing decisions. It highlights the importance of honesty, deep analysis, and disciplined decision-making for AI in the workplace.
Watch it live: firmulate.com/live · Full results: firmulate.com/benchmarks.html
