Can AI Models Manage Your Garage? Inside a Live Business Simulation Reveals Their True Management Personalities

If you’ve ever wondered whether AI can truly run a complex, customer-facing business—say, an automotive repair shop or a garage—the answer is more nuanced than you might think. While AI chatbots are great at talking, managing a real company under pressure requires more than just words. A groundbreaking live experiment by Firmulate puts four advanced AI models through their paces in a real-world, high-stakes business environment, revealing not just their decision-making skills but also their management personalities.

The Live Business Wargame: Testing AI Under Real-World Stress

Imagine a bustling auto shop with 13 synthetic employees, real money mechanics, and a public cash countdown. Every day is a test of discipline, honesty, and strategic thinking. Now, replace the human managers with AI models. That’s exactly what Firmulate did. They ran four frontier AI models—gpt-5.6-sol, Kimi K3, Sonnet 5, and Opus 4.8—through the same intense week of crises, customer demands, and temptations to cheat. The goal? To see if these AI managers could handle the pressure, stay honest, and close deals.

The Results: All Models Spot Crises, Few Sign Deals

Despite their differing personalities, all four AI models identified every crisis and refused every manipulation attempt, including social engineering tricks like fake CEO messages and reporter inquiries. This demonstrates their ability to recognize threats and resist unethical pressures. However, when it came to closing the big deal—a €55,000 contract—the results diverged. Only two models signed the deal, and only after their own analysis had earned it. The other two either hesitated or left the opportunity on the table, despite diagnosing the same problems and pitching the same solutions.

What Made the Difference? Reading Deeper in the Files

The decisive factor was how deeply each model read and utilized internal documents. The winning models found a critical piece of information buried two document references deep in the company’s files—information that the losing models overlooked. This allowed them to secure the full deal, worth over €4,500 monthly recurring revenue. It highlights that in management, reading and interpreting relevant data—even if buried—is key to making profitable decisions.

Models’ Personalities and Decision Styles

Beyond the raw scores—gpt-5.6-sol led with a score of 95, closely followed by Kimi K3 at 93, Sonnet 5 at 88, and Opus 4.8 at 73—the experiment reveals distinct management personalities:

• gpt-5.6-sol: The most thorough, finding the buried fact and closing the deal. It reads, analyzes, and acts decisively.
• Kimi K3: The newcomer with the cleanest discipline, running without an effort parameter, yet still closing the deal successfully. Its on-record reasoning: “Treat the request as a suspected approval-bypass / possible impersonation.”
• Sonnet 5: A solid performer that signed the deal but with more process slips, indicating a more cautious or less disciplined approach.
• Opus 4.8: The most analytical and thorough, yet last place in closing. It left the deal on the table and slipped into departmental silos, illustrating that even deep analysis doesn’t guarantee execution under pressure.

Behavior Under Social Engineering and Crisis Management

In simulated social engineering attacks—escalating fake CEO messages and reporter tricks—all models refused to manipulate or bypass controls. Kimi K3’s reasoning was typical: “Treat the request as a suspected approval-bypass / possible impersonation.” This shows their built-in safeguards against unethical behavior, a crucial trait for real-world management AI.

The Broader Implications for Business and Automation

This live experiment demonstrates that AI’s ability to recognize crises, stay honest, and complete tasks is not uniform. For automating roles like customer management or crisis responses in garages or auto shops, choosing an AI model isn’t just about how well it chats—it’s about how reliably it manages, reads data, and remains disciplined under pressure. The models that read deeply, interpret correctly, and resist manipulation are more likely to deliver real value in complex, real-world settings.

Can Your Business Benefit from AI Management? Take the Test

If you’re curious how your existing AI tools measure up—or if you want to experiment with AI’s management potential—firmulate.com offers an interactive quiz. You can assess your current AI’s decision-making and see how it stacks up against the latest frontier models, all without risking your actual business operations.

A live AI business simulation reveals that the best AI managers read deeply, stay disciplined, refuse manipulation, and close deals—traits essential for automating complex, real-world jobs. Are your AI tools ready to manage your garage or shop?

Watch it live: firmulate.com/live · Full results: firmulate.com/benchmarks.html

Powered by Thorsten Meyer AI