Imagine managing a small interior design firm where every decision, crisis, and temptation is publicly played out in real time — not by humans, but by AI models. This isn’t science fiction; it’s the live experiment from Firmulate, where AI manages a real company, day after day, facing the same challenges you might encounter when choosing new furniture or decor, but with a twist: the stakes are financial, and the decisions are transparent.
The Live Company: AI at the Helm
At the heart of this groundbreaking experiment is a small software company run entirely by AI models that simulate real-world management decisions. It’s not a game; it’s a real-time, publicly auditable experiment that runs every business day at firmulate.com/live. The company has 13 synthetic employees and handles genuine money mechanics, burning €105,000 each month against a modest €2,300 in monthly recurring revenue (MRR).
What makes this experiment extraordinary isn’t just the operational mechanics — it’s the transparency. Every decision, every crisis, and every ethical dilemma is versioned and made available for review. This allows observers to analyze how AI models handle complex scenarios that, in the real world, could mean the difference between success and failure for a business.
AI Builders: Making The Decisions That Turn AI Code Into Real Software
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
The AI Models and Their Performance
Four leading AI models participated in the same week of management chaos: a test to see if they can identify crises, resist manipulation, and close deals without human interference. Their scores out of 100 were impressive: gpt-5.6-sol scored 95, Kimi K3 scored 93, Sonnet 5 scored 88, and Opus 4.8 scored 77. But the real story isn’t just about scores — it’s about what they did, or didn’t, do under pressure.
Remarkably, all models identified every crisis and refused every attempt at social engineering — fake CEO messages escalated over three stages, and a reporter trick asking for a simple yes/no on background were all refused. The models showed integrity, a crucial trait for AI systems that might one day handle sensitive customer data or business operations.
The AI-Driven Leader: Harnessing AI to Make Faster, Smarter Decisions
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
The Hidden Weakness: Reading the Files
While all models performed well on the surface, the real gap was buried two document references deep within the company’s internal files. The models that read these hidden references managed to clinch the deal, closing a €55,000 contract that would have added €4,583 in monthly recurring revenue. This demonstrates that the ability to dig into internal documents — a skill that’s often overlooked — can be decisive in AI-driven decision making.
As an affiliate, we earn on qualifying purchases.
What About Trust and Discipline?
Not all models maintained discipline equally. Opus 4.8, the most thorough participant with over 80 learned rules and the deepest analyses, ended the week at the bottom of the performance league. Its weakness? It left a close opportunity on the table and failed to escalate certain issues, turning discipline into a vulnerability. Meanwhile, the Kimi K3 model, which ran without an effort parameter (a default setting), managed to close the deal with the cleanest discipline of the group.
Successful Construction Project Management: The Practical Guide
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Implications for Interior Design and Beyond
This experiment underscores a crucial point for any business — whether you’re managing a furniture store, an interior design consultancy, or a decor firm: The value of AI isn’t just in generating creative ideas or writing charming emails. It’s in whether an AI can finish what it starts, make honest decisions under pressure, and truly understand your internal documents and strategies.
For interior businesses considering AI tools, this live experiment highlights the importance of transparency, discipline, and the ability to read and interpret internal information. As AI models become more integrated into operational decisions, understanding their limitations and strengths can be the difference between a thriving business and one that struggles to survive.
Experience It Yourself
You can watch this ongoing story unfold at firmulate.com/live. See every decision, read the detailed analysis, and even test your own judgment in the “guess the model” quiz at firmulate.com/quiz.html. For companies eager to evaluate their AI workforce before hiring, there’s also a read-only export option to run your own wargame against your business, without risking real-world consequences.
This live experiment reveals that AI’s true potential in business isn’t just about talent or creativity but about integrity, thoroughness, and discipline. As AI models face real crises and temptations, their ability to read internal data and resist manipulation could define your company’s future — for better or worse.
Watch it live: firmulate.com/live · Full results: firmulate.com/benchmarks.html
