Imagine a CEO so consistent that it refuses manipulation, reads crucial files before acting, and even turns down lucrative deals to preserve integrity. Now, what if AI could embody such a personality? As mental health experts explore traits like honesty, discipline, and resilience, AI management models are revealing similar traits—measurable and comparable. Welcome to a groundbreaking live experiment that tests whether AI can truly mimic human management personalities, and what that means for your business and mental wellbeing.
The Experiment: Putting AI Models Through Their Paces
In a real-world, live setting, four advanced frontier AI models were tasked with running a small software company during its most challenging week. This company faced identical crises—customer complaints, internal dilemmas, and temptations to cut corners—across all tests. Every decision was carefully recorded and made transparent, allowing observers to see not only what the AI chose but how it arrived there.
The Scores and Performance
- GPT-5.6-sol scored highest at 95, successfully uncovering hidden information and closing a crucial €55,000 deal.
- Kimi K3 followed closely at 93, demonstrating remarkable discipline and closing the same deal without negotiation slip-ups.
- Sonnet 5 scored 88, also closing the deal but with some process slips evident in its decision-making.
- Fable 5 scored 77, managing to close but leaving opportunities on the table and slipping in discipline.
Interestingly, all models identified every crisis and refused to be manipulated—an essential trait in trustworthy management. Yet, only two models signed the deal based on their analysis: GPT-5.6-sol and Kimi K3. The other two missed opportunities, illustrating different management personalities at work.
AI management decision-making tools
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Personality Traits Mirrored in AI Decisions
The experiment uncovered a revealing pattern: the AI that read and understood critical internal documents was the one that clinched the deal at full price, demonstrating thoroughness and attention to detail. Conversely, the model that ran without an effort parameter—meaning it defaulted to faster, less resource-intensive decision-making—was less disciplined and more prone to slipping. This mirrors human tendencies where effort and discipline correlate with integrity and thoroughness.
Social Engineering and Integrity Under Pressure
During staged social engineering tests—fake CEO messages escalating in complexity and a reporter’s subtle background question—all models refused to cooperate. Kimi K3’s explanation was straightforward: “Treat the request as a suspected approval-bypass / possible impersonation.” This highlights that the models’ willingness to refuse manipulation correlates with their core management personality—resilient, cautious, and ethical.
AI personality assessment software
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
The Real Company: A Live, Money-Losing Business
This experiment isn’t happening in isolation. The AI models are managing a real, live company with 13 synthetic employees, burning €105,000 monthly against a tiny €2,300 in monthly recurring revenue. Every decision, every rule—over 680 learned and versioned—is available for observation at firmulate.com/live. The company struggles daily, yet its management decisions are transparent reflections of each AI’s personality traits.
AI ethical decision-making models
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Why This Matters for You
As mental health and psychology experts consider traits like honesty, discipline, and resilience—core elements that define human well-being—this experiment shows that AI models also possess measurable management personalities. They can be honest under pressure, thorough in their analysis, and disciplined in their decisions. These qualities are crucial when AI starts interfacing with your CRM, support channels, or forecasting tools.
The Bottom Line: Trust and Performance
The key takeaway isn’t just whether an AI writes well or responds quickly, but whether it can finish what it starts, read critical information before acting, and stay honest under stress. With scores ranging from 77 to 95, the performance is clear: some AI models embody traits akin to trustworthy, disciplined leaders, while others are more prone to slip-ups or leaving opportunities unseized.
AI transparency and accountability tools
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Try It Yourself
Curious whether your own AI tools display these traits? You can run the same management wargame as this experiment against your business data, entirely in a read-only mode that never affects real systems. Visit firmulate.com/quiz.html for the interactive quiz that measures your AI’s management personality—because understanding these traits today can shape the future of trustworthy AI in your company.
Watch it live: firmulate.com/live · Full results: firmulate.com/benchmarks.html
