Get home appliances delivered free with Prime
- Fast, free delivery on millions of items
- Prime Video, Amazon Music and more included
- Member-only deals all year
Can AI Manage a Business Under Pressure? Watch This One Try.
Imagine your smart home devices or appliances not just responding to commands, but actually running a business—facing crises, making decisions, and even losing money every day. Sounds like science fiction? Not anymore. A groundbreaking live experiment puts AI models in the driver’s seat of a real, functioning software company, openly chronicling its daily struggles and decisions. This is AI in the wild—raw, unfiltered, and fascinatingly real.
Products Worth Considering
Ring Alarm 8-Piece Kit (2nd Gen) Home Security System
Ring Alarm 14-Piece Wireless Security System Kit
Philips Hue Festavia 65-Foot Smart String Lights, 250 Color-Changing LEDs
As an affiliate, we earn on qualifying purchases.
The Live Company: An AI-Driven Business in Action
At the heart of this experiment is a company with no human employees—just 13 synthetic “workers” powered by AI models. Every workday, this virtual team faces the same set of crises, decisions, and opportunities, all in a controlled but real-money environment. They operate under over 680 rules they’ve learned themselves, making decisions that impact €105,000 in monthly expenses against a modest €2,300 monthly revenue. This ongoing test is not just a demonstration but a transparent window into how AI manages complex business scenarios in real time, visible at firmulate.com/live.html.
How Do Different AI Models Perform?
Four of the latest frontier AI models were tested in this high-stakes simulation, each running the same week of crises—customer complaints, internal mistakes, manipulative attempts, and even social engineering tricks like fake CEO messages. The models identified every crisis and refused every manipulation, showing integrity at every turn. Yet, only two managed to close a deal worth €55,000—a key measure of success in this experiment—despite all of them diagnosing well and pitching convincingly.
The real breakthrough was lurking in the company’s own files: the models that read and understood the internal documentation could spot a hidden opportunity that others missed, sealing a deal at full price—a remarkable €4,583 in monthly recurring revenue.
Lessons in Business Integrity and Decision-Making
This experiment highlights something critical for businesses considering AI: the importance of honesty and thoroughness. When faced with social engineering, such as a staged “just one yes/no” approval request, all models refused. Kimi K3, one of the models, explained it clearly: “Treat the request as a suspected approval-bypass / possible impersonation.”
The Cost of Running AI in a Business
It’s important to note that this company is not profitable—burning €105,000 each month against a tiny €2,300 revenue. The public cash countdown adds a layer of tension, emphasizing that this is a real, ongoing struggle, not just a demo or a game. The company’s daily operations are versioned, and each decision is auditable, providing a transparent look at the AI’s capabilities and limitations.
What Does This Mean for Your Smart Home?
While this is a far cry from managing a household, the core message is relevant: as AI begins to integrate more deeply into your appliances and home systems, its ability to finish what it starts, stay honest, and handle crises will be crucial. The question isn’t just whether it writes well or responds politely; it’s whether it can make sound decisions under pressure, read your internal files, and keep trust intact—just like these virtual employees attempting to run a business.
The League Table and Final Scores
- gpt-5.6-sol: scored 95, found the buried fact, closed the deal—full performance
- Kimi K3: scored 93, closed the deal with the cleanest discipline
- Sonnet 5: scored 88, closed with minor slips
- Fable 5: scored 77, closed but with process weaknesses
All models demonstrated they could identify crises and refuse manipulation, but the true test was closing a deal at full value. The experiment’s results are publicly available and ongoing, offering a real-world glimpse into AI’s potential and current limits in business decision-making.
Key Takeaway: Trust, Integrity, and Performance Matter More Than Ever
This experiment proves that AI models can recognize crises and uphold integrity in complex scenarios. However, translating that into profitable, reliable business performance remains a challenge. For consumers and businesses alike, the real question is whether AI can stay honest, read critical internal information, and complete tasks under pressure—whether in running a virtual company or powering your smart home devices. Watching this live experiment unfold offers valuable lessons on the future role of AI in our everyday systems.
Watch it live: firmulate.com/live · Full results: firmulate.com/benchmarks.html
