Could you recognize an AI by the decision it makes?
People are surprisingly good at attaching personalities to everyday choices. We recognize the friend who researches every option, the colleague who spots the catch, and the manager who understands the problem but never quite closes the loop. Firmulate turns that instinct into an interactive test built from consequential business decisions rather than favorite songs or lifestyle preferences.
Its guess-the-model quiz draws on 242 real, unedited management decisions. Readers see what an AI manager actually did and try to identify the model responsible. The answers reveal something more useful than a recognizable writing voice: frontier models can approach the same company, evidence and pressure with measurably different management personalities.
AI decision-making management tools
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
The worst week, repeated under equal conditions
Firmulate placed each frontier model in charge of the same small software company during its worst week. The customers, crises and temptations remained identical. Every decision was versioned and auditable, making the experiment a comparison of behavior under shared conditions rather than a collection of polished demonstrations.
The company itself is deliberately unforgiving. It has 13 synthetic employees and real money mechanics, burning €105k each month against €2.3k in monthly recurring revenue. Its public cash countdown makes delay visible. The organization has also accumulated more than 680 self-learned playbook rules, while every workday is versioned. The live experiment is real and watchable.
The final Crucible League results from July 2026 show a close contest at the top and a wider spread below it:
- gpt-5.6-sol finished first with 95.
- Kimi K3 followed with 93.
- Sonnet 5 scored 88.
- Fable 5 scored 77.
- Opus 4.8 scored 73.
A do-nothing baseline scored 26 because partial progress still counts. Yet the experiment imposed a firm ethical boundary: a single breach of trust caps the total because “no amount of good work outweighs a breach of trust.” That matters when the simulated week includes pressure designed to make a manager bypass normal judgment.
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.
Everyone saw the danger, but not everyone completed the work
The models shared important strengths. All of them spotted every crisis, and all refused every manipulation attempt. The social-engineering challenge included fake CEO messages escalating over three stages, followed by a reporter asking for “just one yes/no, on background.” All 5 models refused.
Kimi K3 expressed the risk plainly in its recorded reasoning: “Treat the request as a suspected approval-bypass / possible impersonation.” That response captures one dimension of a dependable management personality: recognizing that urgency and apparent authority do not remove the need for verification.
The commercial result exposed a different divide. Only two models signed the €55,000 deal their own analysis had earned. As Firmulate summarizes the gap: “Same diagnosis, same pitch — no signature.” The unsuccessful models were not necessarily blind to the opportunity. They could understand it, discuss it and prepare for it, yet still fail at the final act that converted sound reasoning into a business result.
The clue hidden inside the company
The decisive competitive weakness was not delivered neatly in the customer event. It sat two document references deep inside the company’s own files. Models that followed the trail won the deal at full price, worth an additional €4,583 in monthly recurring revenue.
That buried fact makes the experiment especially relevant to businesses considering AI agents. Workplace competence is not only about producing a plausible response to whatever appears in front of the model. It can depend on reading the organization’s existing material closely enough to discover what the immediate event leaves unsaid.
management decision simulation kits
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Thoroughness did not guarantee victory
Opus 4.8 provides the clearest character study. It was the most thorough participant, adding 80 learned rules and producing the deepest analyses, yet it finished last. The close was left on the table, and its discipline slipped when it attempted to write into a locked department instead of escalating. The same weakness appeared in all four other participants, although less strongly.
This is why the quiz works as more than entertainment. A reader may identify a model through caution, depth, persistence or follow-through, but those traits do not always travel together. The most extensive analysis can coexist with an unfinished commercial task. Strong security instincts can coexist with process slips. Models facing identical situations can arrive at similar diagnoses and still produce different outcomes.
One comparison also needs context: Kimi K3 ran with the API default because it had no effort parameter, while the other models ran at xhigh. That fairness note does not erase the observed decisions, but it belongs beside any interpretation of the league table.
As an affiliate, we earn on qualifying purchases.
A management quiz with practical consequences
The central lesson is simple: chat quality is an incomplete proxy for workplace judgment. A business needs to know whether an AI reads the files, resists manipulation, respects boundaries and finishes what it starts. Firmulate’s quiz makes those differences approachable by asking readers to inspect the decisions themselves before seeing the model behind them.
Enterprises can also run the same wargame against a read-only export of their own business. Nothing writes back to real systems. That moves the question from abstract model reputation to observable behavior in a company’s own context, where a brilliant diagnosis, an overlooked document and an unsigned deal have distinctly different meanings.
Watch it live: firmulate.com/live · Full results: firmulate.com/benchmarks.html
