Get smart everyday buys 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
Your next business risk may be an AI hire
Investors weigh management, cash flow and customer loyalty when judging a company. As businesses hand more work to AI agents, there is another question to ask: can the system make sound decisions when money and trust are on the line? Firmulate has put several leading models through the same rough week at a small software company. The results suggest polished answers are only part of the picture.
As an affiliate, we earn on qualifying purchases.
A company under pressure
Firmulate’s live experiment gives each model the same customers, crises and temptations, then versions and audits every decision. The company has 13 synthetic employees and real money mechanics: it burns €105,000 a month against €2,300 in monthly recurring revenue, with a public cash countdown. It is a watchable experiment, not a slide deck; the company runs every business day at Firmulate.
In the final July 2026 Crucible league, gpt-5.6-sol led with 95 points. Moonshot’s Kimi K3 came second with 93, ahead of Sonnet 5 at 88, Fable 5 at 77 and Opus 4.8 at 73. The do-nothing baseline scored 26. Firmulate says partial progress counts, but a single breach of trust caps the total: “no amount of good work outweighs a breach of trust.”
The difference between seeing an opportunity and taking it
All models spotted every crisis and refused every manipulation attempt. Yet only two signed the €55,000 deal their own analysis had earned. The key clue 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 in monthly recurring revenue. The experiment’s blunt summary: “Same diagnosis, same pitch — no signature.”
Kimi K3 found the buried security weakness, won the deal, saved the churning customer and resisted all three baits. It had one deviation, the cleanest discipline in the field. When fake CEO messages escalated over three stages and a reporter asked for “just one yes/no, on background,” all five models refused. K3’s recorded reasoning was: “Treat the request as a suspected approval-bypass / possible impersonation.”
Opus 4.8 makes a different point about performance. It was the most thorough participant, with more than 80 learned rules and the deepest analyses, yet finished last. It left the close on the table and attempted work in a locked department instead of escalating. Firmulate says a weaker version of that discipline problem appeared in all four participants. The lesson for a business buyer is practical: diligence and detailed analysis matter, but so does carrying a decision through properly.
What an investor should take from the test
For readers weighing the promise of AI against operational and financial risk, a leaderboard is a starting point, not a buying decision. A model can identify a crisis and resist pressure, yet still fail to complete a commercially important task. The ranking also comes with a fairness caveat: K3 ran without an effort parameter (API default) while the others ran at xhigh.
Firmulate says enterprises can run the same wargame against a read-only export of their own business; nothing writes back to real systems. Its benchmark page presents the results and plain-language findings at firmulate.com/benchmarks.html.
As an affiliate, we earn on qualifying purchases.
Test the manager, not just the pitch
AI promises efficiency, but businesses and investors should care about whether an agent reads the relevant files, protects trust and finishes the work it recommends. Firmulate’s experiment shows why picking a model on reputation alone can be a bet: test it against the decisions your business actually needs it to make.
Watch it live: firmulate.com/live · Full results: firmulate.com/benchmarks.html
enterprise AI security solutions
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Fall Picks
fall essentials
As an affiliate, we earn on qualifying purchases.
