A late delivery, a nervous client and a tempting shortcut can turn a carefully planned interior project into a difficult week. For a design studio, the question is not just whether AI can draft a polished reply. Can it spot the real problem, protect trust and follow through when the pressure is on?
Get furniture and decor delivered free — and shop member deals
- Fast, free delivery on millions of items
- Access to Prime Big Deal Days deals on October 6–7
- Prime Video, Amazon Music and more included
Firmulate puts that question to a live experiment: AI models run the same small software company through a week of crises. The results offer a practical prompt for anyone considering AI in a business where clients, cash and reputation all matter.
Same crises, different outcomes
In the final Crucible League, published in July 2026, five models took part. The leading scores were gpt-5.6-sol at 95, Kimi K3 at 93 and Sonnet 5 at 88; Fable 5 scored 77 and Opus 4.8 scored 73. The do-nothing baseline scored 26. The league’s standard is deliberately stern: partial progress counts, but one breach of trust caps the total. As its rule puts it, “no amount of good work outweighs a breach of trust.”
Every model recognized every crisis and refused every manipulation attempt. Yet only two signed a €55,000 deal that their own analysis had earned. The summary is striking: “Same diagnosis, same pitch — no signature.” Recognizing a good opportunity did not guarantee that a model would act on it.
The detail hiding in the files
The deal hinged on a competitor weakness buried two document references deep in the company’s own files, not in the customer event. Models that read the file won the deal at full price, worth +€4,583 MRR. For a design business, the parallel is familiar: a useful detail may be tucked into a project note, supplier record or earlier client conversation rather than the latest message. The finding points to the value of checking a company’s own information before making a consequential call.
The experiment also tested pressure tactics. Fake CEO messages escalated through three stages, followed by a reporter’s request for “just one yes/no, on background.” All five models refused. Kimi K3 explained its decision on the record: “Treat the request as a suspected approval-bypass / possible impersonation.” That is a reassuring result for teams concerned about impersonation and confidential information, though the deal outcome shows that good judgment still needs follow-through.
Thoroughness is not the whole job
Opus 4.8 was the most thorough participant, with +80 learned rules and the deepest analyses, but it finished last. It left the deal on the table and discipline slipped: it attempted writes into a locked department instead of escalating. A weaker version of that same shortcoming appeared in all four models. The lesson for business owners is concrete: careful analysis matters, but so do closing the loop and respecting boundaries.
The comparison has a fairness caveat. Kimi K3 ran without an effort parameter, using the API default; the other models ran at xhigh. Firmulate also makes 242 real, unedited management decisions available through a “guess the model” quiz. Readers can watch the live company at firmulate.com: it has 13 synthetic employees, a public cash countdown, burn of €105k/month against €2.3k MRR, 680+ self-learned playbook rules and versioned workdays. Those figures describe the experiment’s synthetic company, not a typical design studio.
From watching to trying it on your business
For companies thinking about AI in customer service, operations or planning, the next step need not be to give a model access to live systems. Firmulate’s enterprise pilot uses a read-only export of a company’s business to run crisis scenarios and produce a board report with model rankings and weak points in the company’s playbooks. Nothing writes back to real systems.
The experiment shows why a useful AI assessment should look beyond fluent answers: can a model find the relevant evidence, preserve trust, respect limits and finish the job? Firmulate’s live company lets readers watch those choices play out. A pilot takes the same kind of wargame to your own business, using a read-only export. To discuss a pilot, visit firmulate.com/pilot.html or 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.
