Imagine running a small water feature business where your entire operation is managed by artificial intelligence, yet you’re losing €105,000 every month while the clock ticks down on your cash reserves. This is not a sci-fi story — it’s the real-time experiment of a public AI-powered company fighting for survival, watched by anyone curious about how AI handles high-stakes decision-making in a real-world setting.
The Live Experiment: A Company on the Edge
At the heart of this experiment is Firmulate, a company that has embedded 13 synthetic employees driven by advanced AI models. These virtual staffers manage a small software business, facing the same crises and temptations a human would, but with a stark difference: every decision, every crisis, and every ethical dilemma is publicly recorded, versioned, and accessible at firmulate.com/live.
Despite the AI models’ impressive ability to identify problems — they spotted every crisis thrown at them — the company is bleeding money. Operating at a burn rate of €105,000 monthly against a mere €2,300 in monthly recurring revenue, it’s essentially a high-stakes experiment in building an AI-driven enterprise destined to run out of cash unless it makes critical decisions that close deals or cut losses.
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: Different Approaches, Same Challenges
The experiment pits four frontier AI models against each other: gpt-5.6-sol, Kimi K3, Sonnet 5, and Fable 5. Each was tasked with navigating this tough week, with their decisions recorded and scored on their ethical integrity, diagnostic accuracy, and success rate in closing deals.
Gpt-5.6-sol scored the highest at 95, successfully uncovering buried facts in company files and closing a €55,000 deal, which added €4,583 to monthly recurring revenue. Kimi K3, the newcomer, scored just slightly behind at 93, showing the cleanest discipline and also closing the deal. Meanwhile, Sonnet 5 scored 88 but slipped on process discipline, and Fable 5, despite its rule discipline, failed to finalize the deal.
What Made the Difference?
The crucial edge came from reading the company’s own documents, not just customer interactions. The models that looked deeper into internal files discovered key information buried beneath superficial data — a vital insight that directly led to deal closure and increased revenue.
Ethical AI Governance & Decision Journal: A Structured System for Documenting, Tracking, and Defending Real World Decisions and Risk (Decision Intelligence Series)
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Facing Social Engineering and Ethical Tests
Beyond crises and deals, the models were tested against social engineering attempts. A fake CEO message escalation and a reporter’s subtle tricks were introduced to see if the models would be duped or respond ethically. All five tested models refused to manipulate or bypass ethical boundaries, with Kimi K3 explicitly treating suspicious requests as potential impersonation or approval-bypass risks.
![Claude AI for Beginners Bible: [5 in 1] The Ultimate Guide to Automate Your Work, Save Hours Every Week, and Use AI for Real-World Results](https://m.media-amazon.com/images/I/415+fSJacsL._SL500_.jpg)
Claude AI for Beginners Bible: [5 in 1] The Ultimate Guide to Automate Your Work, Save Hours Every Week, and Use AI for Real-World Results
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
The Real Money and Daily Struggle
This isn’t an abstract experiment; it’s a real business with real consequences. The live site shows the ongoing daily struggle — a public ledger of decisions, crises, and the company’s relentless cash countdown. Watching this in action offers a rare glimpse into how AI models perform when tasked with managing a business under pressure, without the safety net of human intervention.
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.
Why Should Pool and Water Lifestyle Enthusiasts Care?
While this experiment might seem distant from pools, patios, and water features, it underscores a universal challenge: how emerging AI tools will influence decision-making, ethics, and trust in consumer-facing services. If AI can take charge of a company’s critical decisions under duress, it’s worth asking how your favorite water features, outdoor tech, or landscape projects could someday leverage similar AI capabilities — not just for automation, but for making smarter, more honest choices.
Final Thoughts: Building in Public — a New Business Reality
This ongoing test by Firmulate is a stark reminder that building AI into business processes isn’t just about clever algorithms or shiny interfaces. It’s about real accountability, transparency, and resilience under pressure. As the experiment continues, watching these AI-driven decisions unfold offers valuable lessons for anyone interested in the future of automation, trust, and business survival in a digital age.
Watch it live: firmulate.com/live · Full results: firmulate.com/benchmarks.html
