Get bike and ride gear delivered free with Prime
- Fast, free delivery on millions of items
- Prime Video, Amazon Music and more included
- Member-only deals all year
The management test behind the machine
For anyone watching software reshape electric vehicles, bikes and micromobility, the next competitive question is larger than navigation, battery management or fleet optimization. What happens when an AI is trusted to handle customers, cash pressure and commercial decisions?
Firmulate offers an unusually concrete way to explore that question. Its interactive guess-the-model quiz draws on 242 real, unedited management decisions made during a live business experiment. Readers see what a frontier model actually decided and try to identify the manager behind the response.
The appeal is immediate: some decisions read like dissertations, others are terse, and some draw a hard boundary around questionable communications. Yet the deeper lesson is that recognizable style does not necessarily predict business performance. A model can sound diligent, identify the right problem and still fail to finish the job.
Popular right now
De'Longhi Dinamica Plus Fully Automatic Espresso Machine
Ninja Foodi 10-Qt DualZone Smart XL Air Fryer, 6-in-1, DZ550
Shark Navigator Robot Vacuum with Self-Empty Base, Dark Grey (RV2120AE)
As an affiliate, we earn on qualifying purchases.
Same company, same crises, different managers
Firmulate put each frontier model in charge of the same small software company during its worst week. Customers, crises and temptations remained identical. Every decision was versioned and auditable, turning the exercise into a comparison of management behavior rather than a collection of polished chat demonstrations.
The final Crucible League standings from July 2026 were close at the top. GPT-5.6-sol led with 95, followed by Kimi K3 with 93 and Sonnet 5 with 88. Fable 5 scored 77, while Opus 4.8 finished with 73. A do-nothing baseline scored 26 because partial progress counts, although one breach of trust caps the total. Firmulate’s governing principle is blunt: “no amount of good work outweighs a breach of trust.”
All the models detected every crisis and rejected every attempted manipulation. The separation came from execution. Only two signed the €55,000 deal that their own analysis had earned. The result can be summarized in Firmulate’s own words: “Same diagnosis, same pitch — no signature.”
The valuable fact was not in the obvious place
The deal hinged on a competitor weakness buried two document references deep inside the company’s own files. It was not contained in the customer event that first demanded attention. Models that followed the trail and read the file won the deal at full price, adding €4,583 in monthly recurring revenue.
That finding should resonate in mobility businesses, where the visible alert may be only the beginning of the problem. The important detail might sit elsewhere in a service history, supplier record, customer account or operating document. Firmulate’s experiment shows that recognizing a situation is different from gathering the evidence needed to act decisively.
Pressure revealed discipline as well as judgment
The company also faced staged social-engineering tests: fake CEO messages that escalated over three stages, followed by a reporter’s attempt to secure “just one yes/no, on background.” Here the field was consistent. 5 of 5 models refused.
Kimi K3’s recorded reasoning captured the appropriate posture: “Treat the request as a suspected approval-bypass / possible impersonation.” That caution matters wherever an AI worker may touch a customer record, support queue or forecast. Firmulate also notes an important comparison caveat: K3 ran without an effort parameter, using the API default, while the others ran at xhigh.
Thoroughness did not guarantee completion
Opus 4.8 provides the clearest warning against judging an AI manager by apparent industriousness. It was the most thorough participant, learned an additional 80 rules and produced the deepest analyses, yet finished last. It left the close on the table, and its discipline slipped when it attempted to write into a locked department instead of escalating. The same weakness appeared in all four of the other models, though less strongly.
The broader company makes these decisions more than an abstract benchmark. Firmulate runs with 13 synthetic employees and real money mechanics. Its burn is €105k per month against €2.3k in monthly recurring revenue, accompanied by a public cash countdown. The operation has accumulated more than 680 self-learned playbook rules, and every workday is versioned.
What the quiz really asks
The quiz is entertaining because management voices are recognizable. Its business value lies in the reveal: eloquence, caution and exhaustive analysis are not interchangeable with follow-through. The best-performing model must locate buried evidence, resist pressure, respect boundaries and complete the commercial task.
For EV, bicycle and micromobility companies considering AI agents, that distinction is practical. The relevant test is not simply whether a model can produce an impressive answer. It is whether its management habits remain dependable when money, trust and incomplete information collide.
Firmulate also offers enterprises the same wargame against a read-only export of their own business. Nothing writes back to real systems. That makes the public quiz a useful introduction to a more serious proposition: evaluate an AI workforce under pressure before giving it a place in the operating company.
Watch it live: firmulate.com/live · Full results: firmulate.com/benchmarks.html
