Three of the country's top business schools just tested frontier AI on the analytical work MBAs are trained to do, and the models scored in the high eighties. If you run a company, that number is coming for you soon, probably in a deck that recommends cutting a layer of analysts. So, it is worth being exact about what it measures, because the exact answer is both more useful and more limited than the headline.
The paper is BusinessCaseBench, from researchers at Wharton, Carnegie Mellon, and Harvard Business School. They drew 615 questions from real business school cases across 18 disciplines, from strategy and finance to leadership and ethics. Each model read a case cold and wrote its analysis. A separate grader then compared that analysis against the reference solution the instructor had written, item by item. The models never saw the reference. From the model's perspective, it faced an open business question with no answer attached and answered it well. Under the main metric, Claude Sonnet 4.6 covered about 88% of what the instructor's solution contained, and GPT-5.4 about 87%.
The answer key did not help the model. It could not see the rubric, did not use it, and produced its analysis based solely on the case, as a consultant would from a brief. From the model's perspective, these were genuinely open questions. So it is fair to expect that a model that writes strong analyses on 615 unseen cases will write a strong analysis on the 616th, your real case.
The problem is that "the model will write a similar analysis" and "you will get a similar result" are distinct claims, and only the first is what the benchmark tested.
A Score Is a Comparison
"The model scored 88%" is not a fact about the model's answer by itself. It is a fact about that answer measured against a standard. Two things had to exist for the score to exist – the analysis the model wrote and the instructor's solution it was checked against. The score is the relationship between them.
Now move that setup to your company. The model can still write the analysis, but the standard it's checked against doesn't exist. The 88% was a measure of how well the answer matched a known good answer.
This is not a word game. It is the difference between "the model is competent" and "you can trust the output." The first concerns the answer. The second concerns checking the answer, which requires a standard. The benchmark supplied the standard. Your hardest decisions do not.
Knowable-But-Hidden Is Not the Same as Unknowable
The tempting reply is that the real world is just the benchmark, with the answer hidden. The model handled hidden answers well; a real decision is just another hidden answer.
But the benchmark's answers were not hidden. They were knowable from the start. The professor had already worked the case. A correct answer existed; the model simply was not shown it. That is a different situation from the one you face when deciding whether to enter a market or restructure a division. There, no correct answer exists yet. No one has written it because the outcome that would settle it is years away, the criteria for "good" are contested by the people in the room, and there is no counterfactual to check the decision against, even after the fact.
A graded case has a knowable answer the model didn't see. A live strategic decision has an unknowable answer that doesn't exist to be seen. A student who scores 88% on a past exam she took blind will likely score about 88% on the next exam. But it does not follow that she will make good venture bets, even though both feel like hard, open-ended judgment, because a venture bet has no marking scheme, then or later. The model is the student. The benchmark is the past exam. Your boardroom is the venture bet.
The benchmark is strong evidence for a real claim: Frontier models are good at producing structured business analysis and are improving quickly. It is not evidence for the claim that the score predicts a good outcome for decisions whose standard must be invented rather than looked up. Inventing that standard and deciding what a good answer to your actual question would even need to contain is the judgment.
That argument stands even if the models are excellent.
The Model Is Fully Right About Half the Time
The researchers scored the answers in two ways. The headline 88% reflects partial credit – how much of the instructor's checklist each answer covered. Then they ran a stricter count. They counted how many questions the answer satisfied the entire checklist, every item, with no gaps. That fell to roughly half.
In cases where a correct answer was knowable, the leading model produced a complete answer about once in two cases. The authors put the point in their own title: These are drafts, not verdicts.
Now extrapolate honestly. If you carry this model into the wild and expect "similar performance," you are also carrying its incompleteness. The typical output is strong yet missing something. In the study, a human holding the instructor's solution caught the missing half. In your firm, if you removed the person who could catch it, the missing half would still be missing, and nothing would catch it.
Grader Didn't Check for What Shouldn't Be There
The grader verifies whether each expected point is present. By construction, it does not scan the answer for confident, invented, or incorrect material that falls outside the checklist. A response can meet the expected points and assert three plausible fabrications and still score well, because the method does not look for fabrications.
In the study, the blind spot is harmless because a grader with the solution ignores the extra material. In your company, it is the whole risk. The fabricated line rides along inside a well-organized analysis, unflagged, and the reader of the study who could catch it is the domain expert, whose strong score seemed to make it optional.
Why 'Drafts, Not Verdicts' Is the Expensive Finding
A draft that is 88% right sounds like a bargain, and sometimes it is. But it moves the work rather than removing it. Obvious things that are garbage you would catch. Verified truth is something you could trust. The strong, incomplete, possibly embellished draft is the expensive case, because it earns your trust on the parts you can see and hides its gaps in the parts you would already know the answer to. Catching what it left out requires someone who knows what a complete answer contains, which is exactly the expertise the draft would look to retire.
So, the generation got cheaper, and the checking did not, and for this kind of work the checking cannot be sampled down. You cannot review only the flawed answers, because the flawed ones are the ones that look fine. You review all of them, or you review none and call it oversight. An organization that reads the 88% as a license to remove the reviewers has not automated the analysis. It has removed its own ability to tell when the analysis is wrong.
What It Means If You Run Something
Read correctly, the benchmark is good news. Frontier models are genuinely strong at producing structured business analysis, and that is real leverage you should use. The error is interpreting a score against a known standard as readiness to deploy without a standard.
Two questions to ask of any AI you are about to trust with judgment tasks.
First, does this task have a knowable answer, or is defining the answer the actual job? Where a defensible right answer exists, with a standard you could write down, let the model run and expect it to perform in the wild about as it did on the bench. That extrapolation is fair and worth taking. Where the standard itself must be invented and argued, the model can draft, but a person still owns the decision.
And second, who checks the drafts, and do they know enough to catch what a confident draft leaves out? If the answer is nobody, or nobody who could tell, the high score measures something you will not actually get.
The models are good at the framed question. Your hardest problems arrive unframed. The competitive advantage was never in answering the case. It was in knowing which case you were actually in.
