Imagine a company with no human employees, yet constantly battling daily crises, making tough decisions, and even losing money every single day — all live for the world to see. This is not science fiction; it’s the extreme, build-in-public experiment by Firmulate, revealing how AI models perform under real-world pressures, and what it means for the future of work and business automation.
The Live Company: An Unprecedented Experiment
At the heart of this experiment is a small, simulated software business managed entirely by AI models. The company has 13 synthetic employees, each driven by sophisticated AI, with real money mechanics behind the scenes. Every workday, the company faces the same crises, customer issues, and temptations that real businesses encounter — and all of this is publicly visible at firmulate.com/live.
Currently, the company burns through €105,000 a month while generating only €2,300 in monthly recurring revenue. The public cash countdown adds urgency, and every decision made by the AI is versioned and auditable, providing a transparent window into its decision-making process.
AI management decision-making software
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
How AI Models Perform Under Pressure
The experiment pits four frontier AI models against each other — models designed to simulate management decision-making in crisis scenarios. These models are tested on the same week of adversity, with identical customer issues and crises to see how each handles them. The key findings reveal that:
- All four models successfully identified every crisis, showing strong situational awareness.
- All models refused manipulative or social engineering attacks — even escalation attempts by a fake CEO or a journalist asking for quick approvals.
- Only two of the four models signed the €55,000 deal their own analysis indicated was right. The others, despite diagnosing correctly, left the deal unexecuted, reflecting discipline lapses.
Interestingly, the decisive factor was buried in internal documents: the models that read and understood the company’s own files closed the deal at full price — adding over €4,500 in monthly recurring revenue. Those that missed this key detail failed to complete the sale, highlighting the importance of thorough information processing.
business automation AI tools
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Built-in Public — A New Benchmark in AI Transparency
This experiment isn’t just about AI performance; it’s about transparency and building in public. Every decision, every rule, every version of the AI’s response is documented and accessible in real-time. This level of openness allows anyone to observe how AI agents handle real-world complexities, decision-making under pressure, and ethical considerations.
For example, when faced with a social engineering scenario—such as a request from a fake CEO or a reporter asking for a quick approval—the AIs uniformly refused, citing suspicion and risk. Kimi K3, one of the models, explicitly reasoned: “Treat the request as a suspected approval-bypass / possible impersonation.” This disciplined response underscores a crucial aspect of AI business applications: integrity and honesty matter, especially when stakes are high.
AI transparency and decision tracking software
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Implications for Business and Automation
This live experiment offers a sobering look at the current state of AI in management roles. Despite their ability to diagnose crises accurately, not all models execute decisions reliably. The Opus 4.8 profile, which performed the deepest analysis, ultimately left a deal on the table due to a discipline slip — writing attempts into a locked department instead of escalating them properly.
What does this mean for companies considering AI automation? It’s not enough for AI to understand the situation; it must also act decisively, ethically, and consistently. The question for businesses is: Will your AI perform as expected when faced with real pressure?
AI crisis management simulation
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
The Future of AI in Business — Watch in Real-Time
What makes this experiment groundbreaking is its transparency and live nature. You can watch the company in action at firmulate.com/live, see how different models handle crises, and understand their strengths and weaknesses firsthand. It’s a front-row seat to the evolution of AI-enabled management, where the key performance metrics are integrity, thoroughness, and discipline — not just chat quality or superficial performance.
Furthermore, the platform offers a quiz at firmulate.com/quiz to challenge your management intuition — a way to gauge how well you can predict AI behavior in complex scenarios. For those interested in their own business, the company offers a pilot program to run similar simulations against your own enterprise, with no impact on your actual systems.
Why You Should Care
As AI continues to integrate into core business functions—support, CRM, forecasting—the question isn’t whether AI can write well, but whether it can finish what it starts, stay honest under pressure, and read critical internal data first. The Firmulate experiment highlights that AI’s readiness isn’t just about intelligence; it’s about discipline, transparency, and reliability.
In a world where AI might soon be managing more than just customer chats, understanding its true capabilities and limitations is vital. Watching this live company fight for survival offers a rare glimpse into the future of AI-driven management—where every decision is transparent, auditable, and under public scrutiny.
Watch it live: firmulate.com/live · Full results: firmulate.com/benchmarks.html
