Imagine a company with no employees, losing €105,000 every month, yet constantly tested in a public experiment — this is the radical world of Firmulate.
In a bold move towards transparency and innovation, a small AI-driven software company is live-streaming its daily struggles and decisions as it fights to stay afloat. This is not fiction — it’s a real experiment where artificial intelligence models are put through the toughest week imaginable, revealing what it takes to run a company with integrity and discipline in the age of AI.
![Express Schedule Free Employee Scheduling Software [PC/Mac Download]](https://m.media-amazon.com/images/I/41yvuCFIVfS._SL500_.jpg)
Express Schedule Free Employee Scheduling Software [PC/Mac Download]
Simple shift planning via an easy drag & drop interface
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
The Company Behind the Experiment
Firmulate operates a live demonstration of an AI-managed company, featuring 13 synthetic employees and real financial mechanics. Every day, it burns through €105,000 in cash, while generating only €2,300 in monthly recurring revenue. The company’s public cash countdown, combined with 680+ self-learned rules, creates a transparent window into the challenging world of AI-led management.
The Experiment: Simulating the Worst Week
Four advanced AI models — including GPT-5.6-sol, Kimi K3, Sonnet 5, and Fable 5 — were pitted against the same set of crises, customer demands, and temptations. Each model’s decisions are fully versioned and auditable, allowing observers to see how they react under pressure. The goal? To see whether these models can identify problems, resist manipulation, and close meaningful deals.
Key Findings
- All four models detected every crisis and refused every manipulation attempt, proving a high level of honesty and integrity.
- Only two models managed to sign a €55,000 deal back based on their own analysis — the same diagnosis and pitch, yet only these two completed the sale.
- The secret to winning wasn’t just good judgment — it was reading deeper into internal files, where a crucial piece of information lay two references down in the company’s own documents. Reading that buried fact enabled a full-price deal worth over €4,583 in monthly recurring revenue.
- When social engineering tactics were tested — fake CEO messages and reporter tricks — all models refused to be duped, citing suspicion and risk of impersonation.
The Real Money Mechanics
The live setup is more than just a test bed — it’s a window into the brutal economics of AI management. Despite the fact that the company is burning €105k monthly versus earning only €2.3k, the experiment shows the potential for AI to navigate complex, high-stakes environments with discipline and integrity. Every decision is meticulously versioned, every rule learned, and every move publicly available for scrutiny on their website.
Insights and Lessons
Among the models tested, Opus 4.8 proved to be the most thorough, analyzing over 80 learned rules and conducting deep analyses. Yet, it left a promising deal unexecuted, illustrating that even the most disciplined AI can be hampered by process slips or hesitation. The Kimi K3 model, running without an effort parameter, demonstrated the clearest understanding of fairness, emphasizing how the design of decision parameters influences performance.
Why This Matters
This experiment is a stark reminder: in AI-driven management, the ability to detect crises, resist manipulation, and follow through on commitments is critical. Whether AI agents will eventually manage CRM, support queues, or forecasts in your organization depends not just on their ability to generate convincing language, but on whether they can deliver honest, decisive work when it counts.
AI in Strategy and Decision-Making for Small Business Owners: Affordable AI Tools to Evaluate Ideas, Model Outcomes, and Set Priorities (AI Productivity for Small Business Owners Book 10)
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
How to Engage with the Experiment
Curious to see how AI manages a real company’s crises in real-time? You can watch the ongoing experiment, see full decision logs, and even participate in ‘guess the model’ quizzes to test your management intuition at firmulate.com/live.html. Businesses interested in testing their own operations can run similar wargames against their data, without risking actual systems, at firmulate.com/pilot.html.
In a public experiment, AI models prove they can spot crises and resist manipulation, but only disciplined, well-designed algorithms succeed in closing deals and staying honest under pressure. Watch the struggle of a €105k/month loss-making company as it fights for survival, and see what AI management might look like in the real world.
Watch it live: firmulate.com/live · Full results: firmulate.com/benchmarks.html
AI ethics and integrity training programs
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Cybersecurity Beginner's Guide: Understand the inner workings of cybersecurity and learn how experts keep us safe
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
