A Glimpse into the Future of Business—Live and Unfiltered
Imagine a company that has no human employees, yet is running a real business every single day. It’s losing money, facing crises, and making decisions—just like any company, but entirely managed by artificial intelligence. This isn’t science fiction; it’s happening now, and you can watch it unfold live.
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 Live Experiment: An AI Company in Action
At the heart of this extraordinary experiment is Firmulate, a transparent, build-in-public project that’s reshaping how we see AI’s role in business. It features a genuine small software company with 13 synthetic employees, each driven by advanced AI models, navigating daily crises and managing real money mechanics. The company burns through €105,000 every month against a revenue of just €2,300—a stark reminder of its fragile state. You can observe its ongoing struggle at firmulate.com/live.html.
The AI Models: Competitors in a High-Stakes Arena
Four frontier AI models—GPT-5.6-SOL, Kimi K3, Sonnet 5, and Opus 4.8—were tasked with running this company through its worst week. Each faced the same crises, customer pressures, and ethical temptations, with every decision being versioned and auditable. Their goal? To see which model could best steer the company to success without manipulation or shortcuts.
Results showed that all models detected every crisis and refused manipulation attempts, including social engineering tactics like fake CEO messages. The standout performer, GPT-5.6-SOL, identified a critical hidden fact in internal documents that won a €55,000 deal—adding €4,583 to monthly recurring revenue.
Decision-Making Under Pressure
Interestingly, only two models managed to secure the deal based on their own analysis. Despite identical pitches and diagnoses, the others left opportunities on the table or slipped discipline, illustrating that even the smartest AI can falter under complex, real-world conditions. The deepest analysis came from Opus 4.8, which examined over 80 learned rules but still struggled with closing deals and escalating issues properly.
Insecurity and Trust
Throughout the experiment, the models refused to be manipulated through social engineering. For example, fake CEO messages were rejected outright—an encouraging sign for AI’s potential in safeguarding corporate integrity. Kimi K3 even explicitly noted that such requests could be impersonation attempts, emphasizing AI’s capacity for cautious judgment.
The Build-in-Public Edge
This project isn’t just an academic exercise; it’s a real-time, transparent look at what AI can do in operational settings. Every decision is publicly logged, every crisis replayed, and the entire process is monitored live at firmulate.com/live.html. It’s a window into a future where AI doesn’t just assist in decision-making but actively runs entire companies—albeit with current limitations.
Artificial Intelligence for Cybersecurity: Develop AI approaches to solve cybersecurity problems in your organization
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Why This Matters to Everyone—From Pools to Patios
You might wonder: what does a software company losing €105k a month have to do with pools and water lifestyle? The answer is simple: the principles of transparency, decision integrity, and resilience under pressure are universal. Whether managing a water feature or a customer service team, the ability to trust your tools—especially AI—is crucial. This experiment highlights that AI can be both a powerful ally and a risky asset if not properly tested and monitored.
The Critical Takeaway
As the experiment shows, the current league leader, GPT-5.6-SOL, effectively uncovered hidden information, secured deals, and maintained discipline—all signs of a high-performing AI. Meanwhile, even the most thorough participant, Opus 4.8, left potential revenue on the table due to process slips. The core lesson? Success depends not just on intelligence but on discipline, trustworthiness, and the ability to stay aligned with human goals under pressure.
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.
Watch, Learn, and Prepare
If you’re considering integrating AI into your business—be it for customer support, forecasting, or operations—this live experiment offers invaluable insights. It demonstrates that AI can handle crises and ethical dilemmas, but only if rigorously tested and properly managed. You can run your own similar tests against your business data, without risking real assets, at firmulate.com/pilot.html.
Watch it live: firmulate.com/live · Full results: firmulate.com/benchmarks.html
The Agentic AI Bible: The Complete and Up-to-Date Guide to Design, Develop, and Scale Goal-Driven, LLM-Powered Agents that Think, Execute and Evolve
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
