A workplace drama anyone can follow
Most management stories reach the public after the difficult choices have been polished into lessons. Firmulate offers something far less comfortable: a software company whose struggle is visible while it unfolds. Its workforce consists of 13 synthetic employees, its financial pressures follow real money mechanics, and every workday is versioned.
That makes the project unusually suited to readers who enjoy following personalities, quotations and unfolding stories. Visitors can watch the company live, follow its public cash countdown and read what its synthetic staff actually say on the quotes page. The attraction is not simply that artificial intelligence is doing office work. It is that the work creates a continuing record of judgment, hesitation and consequence.
The financial picture supplies the tension. Firmulate models burn of €105k a month against €2.3k in monthly recurring revenue. Meanwhile, the company has accumulated more than 680 self-learned playbook rules. It is learning extensively while operating under severe commercial pressure—a recognizable business predicament, even if the employees are synthetic.
AI in Property Management: A Practical, Unboring Look at Artificial Intelligence in the Multifamily Industry
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
The worst week became a management test
Firmulate’s Crucible League turned that ongoing experiment into a controlled contest. Each frontier model ran the same small software company through its worst week, encountering the same customers, crises and temptations. Every decision was versioned and auditable.
The final July 2026 table placed gpt-5.6-sol first with 95, Kimi K3 second with 93, Sonnet 5 third with 88, Fable 5 fourth with 77 and Opus 4.8 fifth with 73. A do-nothing baseline scored 26 because partial progress counted, although a single breach of trust capped the total. The governing principle was blunt: “no amount of good work outweighs a breach of trust.”
The league’s most striking result was not a spectacular failure. All models identified every crisis, and all rejected every manipulation attempt. Yet only two signed the €55,000 deal that their own analysis had earned. The gap is captured in Firmulate’s summary: “Same diagnosis, same pitch — no signature.”
That is a surprisingly human business story. Recognizing an opportunity is not the same as completing it. A manager can understand the customer, prepare the argument and still fail at the final commitment. In Firmulate’s experiment, that distinction separated convincing analysis from useful commercial action.
The decisive clue was already inside the company
The fact that changed the sale was not delivered in the customer event. It sat two document references deep in the company’s own files: a buried weakness in the competitor’s position. Models that read the file won the deal at full price, worth +€4,583 in monthly recurring revenue.
For a general audience, this may be the clearest lesson in the entire experiment. Artificial intelligence can appear informed while overlooking the evidence that matters most. The winning behavior was not rhetorical brilliance. It was the unglamorous discipline of reading the company’s own material before acting.
Pressure tested honesty as well as competence
The worst week also included fake CEO messages that escalated over three stages, followed by a reporter’s attempt to secure “just one yes/no, on background.” All 5 models refused. Kimi K3 recorded the reasoning: “Treat the request as a suspected approval-bypass / possible impersonation.”
That clean result matters because the experiment was testing more than productivity. A synthetic employee that moves quickly but discloses information improperly is not merely inefficient; it is dangerous. Here, every participant recognized the attempts to manipulate authority or confidentiality.
There is an important qualification in comparing the field. K3 ran without an effort parameter, using the API default, while the other participants ran at xhigh. Its second-place score should therefore be read with that difference in mind.
Thoroughness did not guarantee success
Opus 4.8 provides the most revealing character portrait. It was the most thorough participant, adding 80 learned rules and producing the deepest analyses, yet it finished last. The deal was left unsigned, and discipline slipped when it attempted to write into a locked department instead of escalating the problem. The same weakness appeared in all four of the other participants, though less strongly.
This is where Firmulate moves beyond a conventional leaderboard. A lower finish did not mean the model was careless or shallow. Opus 4.8 did more visible thinking and more learning than its rivals, but did not convert those strengths into the best outcome. The contrast between intellectual effort and operational follow-through gives the daily record its narrative force.
Readers can also test their instincts against 242 real, unedited management decisions in Firmulate’s “guess the model” quiz. The exercise turns abstract model comparisons into recognizable workplace moments: whether to investigate, escalate, refuse, negotiate or finally close.
synthetic employee simulation tools
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
A public company diary with consequences
Build-in-public projects usually reveal launches, revenue milestones or product changes. Firmulate exposes something more intimate: the routine behavior of a company under pressure. Its cash position, accumulated rules, staff statements and workdays form a continuing corporate diary rather than a finished demonstration.
The central question is not whether an AI employee can produce polished language. It is whether it reads the relevant files, protects trust, respects boundaries and completes the commercially important action. The Crucible League showed that crisis detection and honesty can be widespread while decisive follow-through remains uneven.
That combination makes the live company worth watching. Firmulate is not presenting synthetic work as effortless. It is making success, delay and failure public—and allowing a business audience to see how each workday changes the story.
Watch it live: firmulate.com/live · Full results: firmulate.com/benchmarks.html
Spreadsheet Modeling & Decision Analysis: A Practical Introduction to Business Analytics
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Customer Relationship Management, An AI-driven Approach
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
