Imagine a company with no human employees, losing money every day, yet still living and breathing. What if you could watch it struggle, adapt, and make decisions in real time? This isn’t a sci-fi story — it’s the live experiment by Firmulate, where AI models run a simulated business against harsh realities, and every move is transparent for all to see.
Inside a Living Experiment in AI Management
At the heart of this groundbreaking project is a small software company, operated entirely by artificial intelligence. It’s not just a demo — it’s a fully live, ongoing experiment accessible at firmulate.com/live.html. The company has no employees, no human managers — just 13 synthetic ’employees’ and a complex set of rules guiding their actions. Every decision, crisis management, and strategic move is documented, versioned, and open for public scrutiny.
The goal: test whether AI can truly manage a business, stay honest under pressure, and deliver valuable work — especially when facing its worst week, with same customers, same crises, and the temptation to cheat. Four different AI models, all from the frontier of machine learning, are pitted against this challenge, providing a stark look at what AI can and cannot do today.
The Performance Benchmarks
The experiment measures each model’s ability to diagnose problems, make decisions, and close deals. The top performers, gpt-5.6-sol and Kimi K3, scored 95 and 93 out of 100, respectively, and successfully closed the critical €55,000 deal. In contrast, models like Sonnet 5 and Fable 5 scored 88 and 77, with the latter leaving the deal unexecuted despite identifying it.
What makes these results compelling isn’t just the scores — it’s what they reveal about AI’s decision-making under stress. All models spotted every crisis and refused manipulative or unethical requests, including fake CEO messages and reporter tricks. The only difference: the models that read deeper into company files uncovered a hidden weakness, securing a full-price deal worth an extra €4,583 in monthly recurring revenue.
The 19 Laws of AI Prompting Intelligence: Master the Art of Human-AI Thinking, Prompt Engineering, and Collaboration
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
The Reality of the Live Company
Beyond the scores, this experiment exposes a stark reality: the company burns €105,000 a month while earning only €2,300 in monthly recurring revenue. It’s a fragile, publicly counted-in real time, with every workday versioned and available for review. No fake personas or canned demos here — just a raw display of AI decision-making and company mechanics, accessible at firmulate.com/live.html.
The company operates with over 680 self-learned rules, covering everything from crisis response to strategic negotiations. Even with such detailed guidance, the most thorough AI, Opus 4.8, left a deal on the table — a reminder that discipline and focus remain challenges for AI in complex, real-world scenarios. Interestingly, the AI model K3 ran without an effort parameter (meaning it was more permissive and less disciplined), yet still managed to win the deal by reading the company’s files more thoroughly — an indication that depth of analysis can sometimes outweigh strict rules.
The Lessons for Business and AI
This experiment emphasizes that the question isn’t whether AI can generate convincing chat responses. Instead, it’s whether AI can complete meaningful work under pressure — reading relevant documents, resisting manipulation, and making honest decisions. The real-world implications are profound: if AI is to assist in customer relations, support, or forecasting, it must be capable of trust and reliability, not just language fluency.
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.
Why This Matters to Everyone
The live company’s struggles mirror real business dilemmas: balancing profitability, managing crises, and maintaining integrity. Watching AI models navigate these challenges in a transparent, publicly accessible environment invites business leaders, technologists, and consumers alike to rethink what AI can do — and what it should do.
As the experiment continues, more models are queued to run, and new insights emerge. The underlying message is clear: AI’s capacity to manage complex, ethical, and strategic tasks is advancing — but it’s still a work in progress. The question is no longer if AI can simulate business; it’s whether it can do so honestly and effectively in the real world.
This live experiment showcases AI’s potential and limitations in managing real business challenges — revealing that trust, honesty, and deep analysis are key for AI to truly add value in the workplace.
Watch it live: firmulate.com/live · Full results: firmulate.com/benchmarks.html
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.
AI in Mental Health Nursing: Digital Tools, Clinical Decision Support, Ethics, and Practical Integration Strategies for Psychiatric Nurses and Mental Health Professionals
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
