Imagine if your home renovation assistant could not only suggest the perfect paint shade but also spot hidden flaws in your plans, refuse sneaky shortcuts, and stay honest under pressure. Now, what if that assistant was an AI model? Recent experiments with AI-driven company management are showing us just how different these digital managers can be — and how their personalities could impact your business decisions.
Listen free for 30 days with Audible
Thousands of audiobooks and originals — cancel anytime.
As an affiliate, we earn on qualifying purchases.
Behind the Curtain: The Experiment with AI as Business Managers
At Firmulate, a company specializing in AI management simulations, four advanced AI models were pitted against each other in a controlled environment. Each model was tasked with running a small software company through its worst week — facing the same customer crises, temptations to cut corners, and internal challenges. The goal: see if these AI models could not only navigate the chaos but also make honest, strategic decisions that align with business goals.
The Setup: Real Crises, Same Conditions
The experiment was designed with precision. Every decision was recorded and auditable, and all models faced identical scenarios, including tricky manipulations such as fake CEO messages intended to escalate or bypass approval processes. The company involved was active, with real money mechanics: burning €105,000 monthly against just €2,300 in monthly recurring revenue, illustrating the high-stakes environment.
The Results: Different Personalities, Similar Capabilities
While all four models successfully identified every crisis and refused to fall for manipulation attempts, only two managed to secure the company’s most lucrative deal — a €55,000 contract. Interestingly, the models’ ability to read and analyze critical company documents played a decisive role. Those that identified deeper context within the company files closed the deal at full price, adding approximately €4,583 in monthly recurring revenue.
One standout model, GPT-5.6-sol, scored the highest with 95 points, successfully uncovering the critical information and sealing the deal. Kimi K3, a newcomer, scored a close 93 and demonstrated the cleanest discipline, refusing manipulative tactics without fail. Sonnet 5 followed with an 88, while another derivative scored 77, primarily due to process slips and hesitations during the closing stage.
Personality Profiles of the AI Models
This scoring reveals personality-like traits: GPT-5.6-sol was thorough, digging into the company’s hidden details; Kimi K3 was disciplined and straightforward; Sonnet 5 was somewhat cautious, occasionally slipping on process discipline. Interestingly, the most thorough model, Opus 4.8, scored the lowest, leaving potential deals on the table due to over-analysis and internal communication lapses.
What This Means for Your Business
The takeaway isn’t just about which AI makes the best decisions — it’s about understanding the personalities these models embody. As AI begins to touch your CRM, support queues, or forecasting, the question isn’t merely about chat quality or clever responses. It’s whether the AI can finish what it starts, read your critical documents first, stay honest under pressure, and make strategic choices that benefit your bottom line.
AI decision-making software for business
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
The Human-Like Traits of AI Decision-Making
These experiments highlight that different AI models exhibit management personalities, from meticulousness to discipline to caution. Their ability to detect hidden information and resist manipulative tactics underscores their potential as trustworthy team members or decision-support tools. For example, all models refused to be manipulated through staged CEO messages, citing suspicion of impersonation — a crucial trait in safeguarding business integrity.
Live and Watchable: The Firmulate Company in Action
What’s more compelling is that the company used for this experiment is real and running live, with 13 synthetic employees managing real money mechanics. Every day, the company’s operations are versioned and observable at firmulate.com/live. It serves as a proving ground for AI management, where businesses can simulate their own crises and see how different AI personalities respond — before deploying them in critical roles.
Why Should You Care?
As AI becomes more integrated into daily business operations, understanding its management style and decision-making personality will be vital. Will your AI agent finish what it starts? Will it stay honest under pressure? Will it prioritize your company’s real interests over shortcuts? These are questions the experiment aims to answer, illustrating that the AI’s personality traits can make or break its effectiveness and trustworthiness.
Get Involved: Test Your Business’s AI Management Personality
Interested in seeing how your AI tools stand up to real-world crises? You can run the same management wargame, tailored with your company’s data, at firmulate.com/quiz.html. It’s an interactive way to understand your current AI’s strengths and weaknesses before making critical decisions or investments.
Watch it live: firmulate.com/live · Full results: firmulate.com/benchmarks.html
Flea & tick season Picks
flea and tick prevention
As an affiliate, we earn on qualifying purchases.
