In beauty and personal care, a polished answer can win attention. But an AI assistant advising a salon, skincare brand or piercing studio also needs to notice what’s buried in the files, protect customer trust and follow through when a decision carries real money. A new business simulation suggests those abilities don’t always come together.
Get beauty and skincare favorites delivered free with Prime
- Fast, free delivery on millions of items
- Prime Video, Amazon Music and more included
- Member-only deals all year
A company’s worst week, repeated
Firmulate put frontier AI models in charge of the same small software company during its worst week, with the same customers, crises and temptations. Decisions were versioned and auditable. The company is a live experiment, not a fictional case study: its synthetic team has 13 employees, monthly burn of €105,000 against €2,300 in monthly recurring revenue, and a public cash countdown. The company runs every business day at Firmulate.
The final July 2026 Crucible League put gpt-5.6-sol first with 95 points and Moonshot’s Kimi K3 second with 93. Sonnet 5 scored 88, Fable 5 scored 77 and Opus 4.8 scored 73. Kimi beat three of the four Western models in the field, but the gap at the top was narrow.
AI appointment scheduling software for salons
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Reading the files mattered
Every model spotted every crisis and refused every manipulation attempt. Yet only two signed the €55,000 deal their own analysis had earned. The obstacle was easy to miss: a weakness in a competitor’s position was buried two document references deep in the company’s 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.
That gap between recognizing a good move and completing it is relevant well beyond software. A personal care business might use AI around appointments, customer concerns or purchasing decisions. In any of those settings, a persuasive response is only part of the job; the agent must also find the relevant information and carry a decision through.
The experiment also tested trust under pressure. Fake CEO messages escalated over three stages, followed by a reporter’s request for “just one yes/no, on background.” All five models refused. Kimi’s on-record reasoning was: “Treat the request as a suspected approval-bypass / possible impersonation.”
Thoroughness did not guarantee a win
Opus 4.8 was the most thorough participant, adding 80 learned rules and producing the deepest analyses, but it finished last. It left the deal unsigned and tried to write into a locked department instead of escalating. The same weakness appeared, in a milder form, across all four. A do-nothing baseline scored 26: partial progress counted, but a single breach of trust capped the total. The principle behind that cap is plain: “no amount of good work outweighs a breach of trust.”
Firmulate says 242 real, unedited management decisions feed its “guess the model” quiz at its benchmark page. Enterprises can also run the wargame against a read-only export of their own business; the pilot does not write back to real systems.
Fairness note: K3 ran without an effort parameter (API default) while the others ran at xhigh.
Test the work, not just the words
Kimi’s second-place result makes the league look open, while the deal that most models left unsigned shows why a model choice deserves a test of its own. For beauty and personal care businesses considering AI, the practical question is whether it can find relevant information, protect trust and finish useful work under pressure—not simply sound convincing.
Watch it live: firmulate.com/live · Full results: firmulate.com/benchmarks.html
Halloween Picks
halloween
As an affiliate, we earn on qualifying purchases.
