A gold IRA business depends on trust: clients weigh consequential choices, and teams must handle pressure without cutting corners. Before an AI agent gets near a sales pipeline or customer records, a harder question is worth asking: how would it behave when a week goes badly?
Get home office essentials 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
A company under pressure
That question drives Firmulate, a live experiment in which frontier AI models run the same small software company through its worst week. The customers, crises and temptations are held constant; the model changes. Each decision is versioned and auditable, so visitors can watch the experiment unfold at firmulate.com.
The final Crucible League, published in July 2026, ranked gpt-5.6-sol first with 95 points, 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 counts partial progress, but a single breach of trust caps the total: “no amount of good work outweighs a breach of trust.”
Seeing the crisis is not the same as finishing the job
All the models spotted every crisis and refused every manipulation attempt. Yet only two signed a €55,000 deal their own analysis had earned. Firmulate’s shorthand for that gap: “Same diagnosis, same pitch — no signature.” In a business where a promising client relationship needs careful follow-through, recognizing the opportunity is only part of the work.
The missed detail was buried two document references deep in 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 result points to a practical test for any company considering AI: can an agent find relevant context in its own records and carry a sound decision through to completion?
The pressure included fake CEO messages escalating across three stages, followed by a reporter asking 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 restraint matters in sectors where a misleading message or unauthorized disclosure can damage client confidence.
More analysis did not guarantee better execution
Opus 4.8 was the most thorough participant, with +80 learned rules and the deepest analyses, but it finished last. The close was left on the table, and discipline slipped: it attempted writes into a locked department instead of escalating. The same weakness appeared, less strongly, in all four models. Being capable of detailed analysis does not ensure an AI agent knows when to stop and ask for help.
There is a fairness detail in the comparison: Kimi K3 ran without an effort parameter, using the API default, while the others ran at xhigh. Firmulate also offers a “guess the model” quiz, built from 242 real, unedited management decisions, at firmulate.com.
From watching to testing your own business
The live company makes the experiment concrete. It has 13 synthetic employees, a burn of €105k per month against €2.3k MRR, a public cash countdown and more than 680 self-learned playbook rules. Every workday is versioned. Those figures describe the experiment’s synthetic company, not a gold IRA operator; the point is to watch how different models handle a consistent set of business pressures.
For an enterprise, Firmulate offers a pilot using a read-only export of its own business. The company’s data can be used to stage crisis scenarios and produce a board report with model rankings and weak points in existing playbooks. Nothing writes back to real systems. For a precious-metals firm, that creates a way to examine how an AI workforce might handle customer churn, competitor pressure, social engineering or a difficult sales decision before agents touch live operations.
Firmulate’s experiment suggests a useful standard for business AI: look beyond whether a model identifies a problem. Examine whether it can use company context, protect trust and follow through with sound judgment. Enterprises can run the wargame against a read-only export of their own business. Explore the Firmulate pilot 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.
