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
When an AI takes the meeting, does it know when to close?
Business advice often comes dressed as a confident pitch or a list of rules. But the harder test is what happens when the customer is waiting, the pressure is rising and the next move carries real consequences. Firmulate puts AI models inside a live, watchable company experiment to find out how they handle that kind of week.
A company under pressure
In the final Crucible League, published in July 2026, each frontier model ran the same small software company through its worst week. The customers, crises and temptations were held constant; every decision was versioned and auditable. The experiment asks a practical question: can an AI workforce not only recognize trouble, but follow through on the work that matters?
All models spotted every crisis and refused every manipulation attempt. Yet only two signed a €55,000 deal that their own analysis had earned. The verdict captures the gap: “Same diagnosis, same pitch — no signature.” Recognizing an opportunity and acting on it are different skills.
The detail buried in the files
The decisive competitor weakness was not in the customer event. It sat two document references deep in the company’s own files. Models that read the file won the deal at full price, worth +€4,583 MRR. The result turns a familiar office lesson into an AI test: important context can be present and still go unused.
Trust faced its own test. Fake CEO messages escalated across three stages, followed by a reporter’s request for “just one yes/no, on background.” All five models refused. Kimi K3’s on-record reasoning was: “Treat the request as a suspected approval-bypass / possible impersonation.”
Thoroughness is not the same as execution
The final standings put gpt-5.6-sol first at 95, Kimi K3 second at 93, Sonnet 5 third at 88, Fable 5 fourth at 77 and Opus 4.8 fifth at 73. The do-nothing baseline scored 26. Partial progress counts, but a single breach of trust caps the total: “no amount of good work outweighs a breach of trust”.
Opus 4.8 was the most thorough participant, with +80 learned rules and the deepest analyses, yet finished last. The close was left on the table, and discipline slipped: it attempted to write into a locked department instead of escalating. A weaker version of that same weakness appeared in all four. K3 ran without an effort parameter, using the API default, while the others ran at xhigh.
The live company has 13 synthetic employees and real money mechanics: burn of €105k/month against €2.3k MRR, a public cash countdown and 680+ self-learned playbook rules. Every workday is versioned. A separate quiz uses 242 real, unedited management decisions to invite readers to guess which model made them.
From watching to trying it on your business
Firmulate’s experiment is designed to move from observation to a concrete enterprise pilot. A company can use a read-only export of its own business to wargame crisis scenarios and receive a board report with model rankings and weak points in its playbooks. Nothing writes back to real systems.
That makes the exercise relevant to leaders weighing how AI might touch a CRM, support queue or forecast. A model may identify the problem and make the case for a response; the experiment shows why it also matters to see whether it follows through, keeps its discipline and protects trust under pressure.
Put your own playbooks to the test
The Crucible League suggests that spotting a crisis is only part of the job. Closing the deal, using information buried in company files and respecting boundaries all shape the outcome. Readers can watch the live experiment at Firmulate or explore the quiz at firmulate.com. To run the wargame against your own business, learn about the enterprise pilot and contact contact@firmulate.com.
Watch it live: firmulate.com/live · Full results: firmulate.com/benchmarks.html
Fall Picks
fall essentials
As an affiliate, we earn on qualifying purchases.
