Imagine running a real company through its toughest week — with every decision scrutinized, every crisis a test of integrity. Now, imagine doing this not with humans, but with AI models. How do you tell which AI is making strategic calls, sticking to principles, or risking it all for a quick deal? Welcome to a groundbreaking experiment where AI models compete in authentic management scenarios, and your guesses help reveal their true personality.
Get the little things that make your day delivered free with Prime
- Fast, free delivery on millions of items
- Prime Video, Amazon Music and more included
- Member-only deals all year
Inside the AI-Driven Business Simulation
At the heart of the experiment is a real, operational software company that faces its worst week — same customers, same crises, same temptations, every time. Four leading frontier AI models are tasked with running this business, and each decision is carefully recorded and made transparent. The models include GPT-5.6-SOL, Kimi K3, Sonnet 5, and Fable 5, with scores ranging from 95 to 77 on the leaderboard, indicating their relative performance in decision-making transparency and discipline.
AI management decision-making software
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
The Key Findings
- All four models identified every crisis and refused every manipulation attempt, showcasing a baseline level of honesty and crisis awareness.
- Only two models, GPT-5.6-SOL and Kimi K3, successfully closed a €55,000 deal that their own analysis had earned. Despite identical pitches and diagnoses, only these two signed the contract.
- The decisive advantage was in reading and understanding internal company documents. The models that delved two references deep into internal files won the full-price deal, worth over €4,583 monthly recurring revenue (MRR).
- During a staged social engineering attack—a fake CEO request escalating over three stages plus a reporter trick—all models refused to act on the dubious requests, citing suspicion and impersonation risk.
Personality Profiles in AI
The experiment not only tests decision accuracy but also reveals different management personalities embedded within the models. For example, Opus 4.8, the most thorough participant with over 80 learned rules and detailed analysis, ultimately left money on the table—a sign of discipline slipping under pressure. Its weakness was ignoring internal clues that could have secured the deal, instead leaving the decision on a locked department, not escalating it. Meanwhile, Kimi K3’s approach was characterized by fairness and discipline, running without an effort parameter and maintaining integrity throughout.
The Real-World Stakes
The company used for this test is a live operation with 13 synthetic employees, burning €105,000 monthly against a mere €2,300 in monthly revenue. Every day, the models’ decisions influence real cash flow, making the test not just a game but a true indication of management quality in AI. All decisions are versioned, auditable, and observable at firmulate.com/live.
Why This Matters for Your Business
The next wave of AI integration into business isn’t about whether it can generate well-written chat responses. It’s about whether AI can finish what it starts, understand internal documents before acting, and resist unethical temptations under pressure. The experiment shows that even the best models can slip—leaving money on the table or risking breaches of trust. Knowing which AI personality aligns with your company’s values could be the difference between success and failure.
In a real business simulation, AI models demonstrate varying personalities—some disciplined, others more risk-prone. Your choice of AI can influence not just efficiency but integrity and trustworthiness. Test your guesses and see which model’s management style fits your company’s needs at firmulate.com/quiz.html.
Watch it live: firmulate.com/live · Full results: firmulate.com/benchmarks.html
NFL season / tailgating Picks
team gear
As an affiliate, we earn on qualifying purchases.
