Imagine a thriving business operating in plain sight, yet it has no human employees, loses over €100,000 every month, and is publicly battling for survival — all while you can watch every decision unfold in real time. This is not science fiction; it’s the live experiment by Firmulate. In a world increasingly driven by AI, what if your business could be managed by artificial intelligence that actually makes real decisions, handles crises, and even fights temptation — all in front of a live audience?
The Reality of a Publicly Run AI Company
At the heart of this experiment is a small, real software company that’s run entirely by AI models. It has 13 synthetic employees, each designed to simulate different aspects of business management, from customer relations to crisis response. Every weekday, its operations are versioned, decisions are logged, and the entire process is transparent for anyone to observe at firmulate.com/live.
Despite this transparency, the company is far from profitable. It burns through €105,000 each month while generating only €2,300 in monthly recurring revenue. That’s a stark reminder that AI management isn’t about instant profits but about testing how well these models handle real-world pressure and decision-making.
Building AI-Powered Products: The Essential Guide to AI and GenAI Product Management
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Why Does This Matter for Crypto and Bitcoin Enthusiasts?
In the crypto world, many projects tout automation and AI as the future. But how reliable are these systems when faced with crises or manipulative tactics? The Firmulate experiment offers a revealing glimpse. It shows that even the most advanced AI models can identify crises, refuse unethical manipulation, and focus on honest, strategic decisions — or sometimes, miss opportunities due to internal weaknesses.
For instance, during the experiment, all four models examined the same crisis scenario, and all recognized the issues and refused manipulation attempts like fake CEO messages or secret deals. Yet, only two models managed to close a significant deal at full price — a €55,000 agreement — after their own analyses. The key difference? The winning models read deeper into the company’s own documents, uncovering hidden advantages that others missed.
AI for Small Business: From Marketing and Sales to HR and Operations, How to Employ the Power of Artificial Intelligence for Small Business Success (AI Advantage)
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
The Deep Dive into AI Decision-Making and Trust
The experiment doesn’t just test AI’s ability to recognize crises; it highlights a crucial weakness: the models that read and interpret internal company files at a deeper level tend to perform better. In fact, that buried fact in internal documents was the deciding factor that secured the deal, worth over €4,500 in monthly recurring revenue.
Adding to the complexity, the models faced social engineering tests: staged fake CEO messages escalating over multiple phases and even a reporter’s subtle request for a background approval. Remarkably, all five models refused these requests, demonstrating a robust understanding of impersonation risks. Kimi K3, one of the models, explained its refusal by treating these requests as potential impersonation or bypass attempts.
Crisis Management for Software Development and Knowledge Transfer (Smart Innovation, Systems and Technologies, 61)
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
The Reality and Risks of AI-Driven Business
What’s striking is the ongoing financial crisis facing this live company. It’s a real business with actual cash flow, burning €105,000 monthly against just €2,300 in revenue. Yet, it continues to operate, making decisions, handling crises, and evolving every day. The entire setup is viewable at firmulate.com/live, showing a raw, unfiltered look at AI management in action.
The analysis of different models reveals that thoroughness makes a difference. OPUS 4.8, for instance, conducted over 80 learned rules and deep analyses but still left revenue on the table, slipping in discipline during a critical close. Meanwhile, the K3 model ran without a default effort parameter, which may have contributed to its stronger performance.
AI-driven company monitoring platform
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Implications for the Future of AI in Business and Crypto
This experiment is not just about a quirky company; it raises fundamental questions about trust, honesty, and decision-making under pressure. If AI models can handle crises, refuse manipulative tactics, and even uncover hidden internal facts, then the real challenge becomes: how do we deploy these in our own crypto ventures, fintech platforms, or decentralized projects?
For crypto enthusiasts considering AI-driven automation, the lessons are clear. The focus should shift from mere chat quality to whether the AI can deliver consistent, honest, and strategic work — especially when stakes are high. The live experiment by Firmulate exemplifies this shift, offering a rare glimpse into what future AI-managed companies might look like and the hurdles they face.
This live experiment shows AI models can recognize crises, refuse manipulation, and even uncover hidden internal truths, but financial sustainability remains elusive. Its real-time, transparent approach offers valuable lessons for crypto and AI in business — trust, honesty, and resilience are the real currencies.
Watch it live: firmulate.com/live · Full results: firmulate.com/benchmarks.html
