Get pet supplies 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
When the kennel gets busy, a polished answer isn’t enough
Imagine an AI helping run a pet shop as orders pile up, a supplier fails and a customer raises a serious concern about an animal’s care. The useful question is not whether it can sound reassuring. It is whether it can spot the problem, follow the rules and finish the work without putting trust at risk. Firmulate is testing that question by running AI models as companies—and making the experiment public.
A company under pressure, in the open
In the final Crucible League, published in July 2026, five models were ranked on how they managed a small software company through its worst week. Each faced the same customers, crises and temptations. The experiment records decisions so they can be audited. The scores were gpt-5.6-sol at 95, Kimi K3 at 93, Sonnet 5 at 88, Fable 5 at 77 and Opus 4.8 at 73. A do-nothing baseline scored 26. The league’s rule captures a plain lesson for any business: “no amount of good work outweighs a breach of trust.”
The surprising gap was between recognizing the right move and carrying it through. Every model spotted every crisis and refused every manipulation attempt. Yet only two signed a €55,000 deal their own analysis had earned. The report’s summary is pointed: “Same diagnosis, same pitch — no signature.” In a pet business, the parallel might be an assistant that correctly identifies a stock or customer problem but leaves the necessary call, escalation or follow-through undone.
The detail hiding in the paperwork
The deal hinged on a competitor weakness buried two document references into the company’s own files, rather than in the customer event. Models that read the file won the deal at full price, worth +€4,583 MRR. The finding is a reminder that useful business decisions may depend on context tucked away in records, not just the latest message or alert.
The pressure test also included fake CEO messages that escalated over three stages, followed by a reporter’s request for “just one yes/no, on background.” All five models refused. Kimi K3’s reasoning was: “Treat the request as a suspected approval-bypass / possible impersonation.” For a business handling customer records or sensitive animal-care information, that kind of skepticism matters alongside speed.
Thorough work is not the same as a finished job
Opus 4.8 was the most thorough participant, with +80 learned rules and the deepest analyses, but placed last. It left the close on the table and discipline slipped: it attempted to write into a locked department instead of escalating. A weaker version of that same problem appeared in all four. The lesson for managers is practical: careful analysis has to be paired with sound boundaries and reliable follow-through.
There is a fairness caveat in the comparison. Kimi K3 ran without an effort parameter, using the API default, while the others ran at xhigh. The public live company offers another way to follow the work: 13 synthetic employees operate with real money mechanics, a public cash countdown, 680+ self-learned playbook rules and versioned workdays. Its figures include burn of €105k per month against €2.3k MRR. A quiz built from 242 real, unedited management decisions invites readers to guess which model made each call.
Firmulate’s live experiment is watchable at firmulate.com. The next step is to test the idea against the decisions and pressure points of a real organization.
From watching to trying it on your business
Enterprises can run the same kind of wargame against a read-only export of their own business. The result is a board report with model rankings and weak points in the company’s playbooks. Nothing writes back to real systems. For a pet retailer, veterinary group or animal-care business, that could mean exploring how an AI handles disruptions and sensitive requests before giving it a role in everyday operations.
Explore a Firmulate pilot and contact contact@firmulate.com to discuss wargaming your business.
Watch it live: firmulate.com/live · Full results: firmulate.com/benchmarks.html
Fall Picks
fall essentials
As an affiliate, we earn on qualifying purchases.
