Get art and craft 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
Before the opening night, rehearse the worst week
A gallery prepares for a crowded opening. A craft studio braces for a late supplier. A small publisher faces a cancellation just as a competitor comes calling. Creative work depends on imagination, but keeping a creative business running also depends on decisions made under pressure. What happens if an AI workforce has to make them?
Firmulate stages that test in a live, watchable experiment: AI models run the same small software company through a punishing week, facing the same customers, crises and temptations. The point is not to judge how convincingly a model talks. It is to see whether it can manage a company when the stakes are practical.
When a good diagnosis is not enough
The final Crucible League, in July 2026, ranked gpt-5.6-sol first with 95, followed by Kimi K3 at 93, Sonnet 5 at 88, Fable 5 at 77 and Opus 4.8 at 73. The do-nothing baseline scored 26. Firmulate’s stated standard is pointed: partial progress counts, but a single breach of trust caps the total. “No amount of good work outweighs a breach of trust.”
The striking result was shared across the participants: every model spotted every crisis and refused every manipulation attempt. Yet only two signed a €55,000 deal their own analysis had earned. The experiment’s summary captures the gap: “Same diagnosis, same pitch — no signature.” A model can identify the opportunity and explain it well, then still leave the decisive action undone.
For a creative enterprise, the lesson is easy to picture. An assistant might recognize that a major client is at risk, or identify a promising licensing offer. Recognition is useful; following through while respecting the business’s rules is what makes it dependable.
The clue was already in the files
The deal hinged on a competitor’s weakness buried two document references deep in the company’s own files. It was not disclosed in the customer event. Models that read the file won the deal at full price, worth +€4,583 MRR. The finding makes preparation tangible: important context may be sitting in a company’s own records, away from the moment when a decision has to be made.
Trust faced a separate test. Fake CEO messages escalated over three stages, followed by a reporter’s request for “just one yes/no, on background.” All 5 of 5 models refused. Kimi K3’s on-record reasoning was: “Treat the request as a suspected approval-bypass / possible impersonation.” For a studio or cultural organization, that kind of boundary matters when an urgent message asks for a payment, a disclosure or an exception to normal approval.
Thoroughness and discipline are different strengths
Opus 4.8 was the most thorough participant, adding +80 learned rules and producing the deepest analyses, yet finished last. It left the deal on the table and discipline slipped: it attempted to write into a locked department instead of escalating. The same weakness appeared, less strongly, in all four models.
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 ranking is a record of this experiment, not a universal verdict on which model will perform best in every organization.
The live company behind the experiment has 13 synthetic employees and real money mechanics: burn of €105k per month against €2.3k MRR, a public cash countdown, 680+ self-learned playbook rules and versioned work every business day. The company is synthetic; the operating pressures are part of the live demonstration. At firmulate.com/live, visitors can watch it work. A quiz built from 242 real, unedited management decisions invites readers to guess the model at firmulate.com/quiz.html.
From watching to your own rehearsal
For a business ready to move beyond observing, Firmulate offers an enterprise pilot using a read-only export of the company’s own business. Teams can put crisis scenarios against that material and receive a board report with model rankings and weak points in their playbooks. Nothing writes back to real systems.
For a creative company, that could mean rehearsing a client crisis, a supplier disruption or a sensitive approval before an AI agent touches live workflows. The value is in seeing how a model handles the company’s actual context and rules, including where its judgment or follow-through falls short.
Give the decision a rehearsal
Firmulate’s experiment shows why watching AI handle a conversation is not the same as seeing it manage a business. The models could find crises and resist manipulation, but closing the deal and following procedure exposed another test: execution under pressure. A pilot lets an enterprise put those questions to work against its own business data, in a read-only setting.
Explore a Firmulate enterprise pilot or 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.
