"There is an epidemic failure within the game to understand what is really happening."
When Peter Brand delivers that line in Moneyball, he is talking about baseball.
He could just as easily be talking about artificial intelligence in 2026.
Every day, companies announce new AI initiatives. Vendors race to add AI to product names. Consultants publish AI frameworks. Executives scramble to determine which model, platform, or agent they should deploy.
The conversation sounds remarkably familiar.
In baseball, teams thought they were buying players.
What they were really buying was wins.
The teams that understood the difference changed the game.
Today, many firms believe they are buying AI.
What they should be buying is business outcomes.
The Great AI Misunderstanding
In Moneyball, Peter Brand explains that baseball executives had developed an imperfect understanding of where wins came from.
As a result, they overvalued certain players, undervalued others, and spent money in the wrong places.
The same thing is happening with AI.
Organizations compare models. They compare vendors. They compare features. They compare who has the newest agent framework or the largest context window.
Yet very few stop to ask the most important question:
What business result are we actually trying to create?
- Reducing fund reporting time by 50%.
- Cutting operational costs.
- Improving investor response times.
- Reducing reconciliation breaks.
- Increasing distribution capacity.
- Accelerating onboarding.
These are outcomes.
AI is merely one possible way to achieve them.
The problem, however, runs deeper than technology selection.
Most AI conversations eventually become debates about models, vendors, copilots, agents, and benchmarks. Those discussions matter. But they are still happening at the wrong layer.
The breakthrough in Moneyball was not finding better players.
It was understanding where wins actually came from.
Once Billy Beane understood that getting on base produced runs and runs produced wins, player selection became much easier.
The same principle applies to AI.
Before evaluating models, organizations need to understand where intelligence actually comes from inside their business.
And in private markets, that intelligence rarely lives inside a public LLM.
Companies Are Buying Players Instead of Wins
Many AI projects begin with technology selection.
Should we use OpenAI?
Anthropic?
Google?
Microsoft?
Should we deploy agents?
Copilots?
Custom GPTs?
The thinking is almost identical to a baseball team debating which star player to acquire.
The technology becomes the center of the discussion.
The outcome becomes secondary.
This often leads to expensive pilot programs that generate excitement but little measurable value. The organization ends up with more AI tools but no meaningful improvement in operating performance.
The scoreboard remains unchanged.
A more important question is rarely asked:
Where does the expertise behind our business processes actually live?
For private markets firms, the answer is almost never "inside the model."
It lives in the business itself.
It lives in fund accounting procedures. It lives in investor reporting workflows. It lives in onboarding processes, compliance reviews, valuation committees, capital activity processing, and the thousands of decisions experienced operators make every reporting cycle.
A public LLM can explain what a capital call is.
It can summarize an ILPA reporting standard.
It can draft an investor communication.
What it cannot do is understand how your organization handles commitments across feeder funds, parallel vehicles, co-investments, and side-letter arrangements.
It does not know why your operations team reviews certain balances before investor reports are released.
It does not know which exceptions require escalation, which require CFO approval, and which can be resolved automatically.
That knowledge exists inside the business.
It exists in procedures, spreadsheets, workflows, approval chains, and years of accumulated institutional experience.
That operational intelligence is often far more valuable than the model itself.
Why Smart Companies Still Get This Wrong
If this sounds obvious, why do so many intelligent organizations continue making the same mistake?
Because models are visible.
Vendors can demonstrate them.
Analysts can benchmark them.
Consultants can create scorecards around them.
Boards can discuss them.
Operational intelligence is much harder to see.
There is no benchmark for twenty years of fund accounting expertise.
There is no leaderboard for institutional knowledge embedded inside a reporting team.
There is no Gartner quadrant for how an operations group manages reporting exceptions during quarter-end.
Yet those assets often create more value than the technology being evaluated.
As a result, firms gravitate toward the visible thing even when it is not the source of the advantage.
That is exactly the mistake baseball made.
The AI Procurement Process Is Backwards
Most firms follow a process that looks something like this:
The Technology-First Trap
Select Vendor → Deploy AI → Search for Use Cases → Hope for Results
The Moneyball Approach
Define Outcome → Map Workflow → Identify Operational Intelligence → Select Technology → Measure Results
Notice that model selection does not disappear.
It simply moves to the end of the process where it belongs.
The goal is not to find the smartest model.
The goal is to understand where intelligence already exists inside the business and determine how AI can amplify it.
The Private Markets Example
Consider investor inquiries. Many firms immediately jump to the idea of deploying an AI-powered chatbot to answer LP questions. That may be the right answer, but it may also be solving the wrong problem. Before selecting a tool, firms should ask why investors are asking those questions in the first place. Are capital account statements difficult to interpret? Is information inconsistent between the investor portal, quarterly reports, and fund administrator records? Are reporting packages arriving later than expected? Do investors need additional context around valuations, distributions, or capital calls? In many cases, the root issue is not communication. It is operational intelligence. An AI chatbot may answer questions faster, but improving the underlying reporting process, data quality, and operational workflows may eliminate many of those questions altogether. That distinction matters. One approach automates activity. The other improves the operating model.
The Firms That Win Will Think Differently
The organizations that benefit most from AI will not necessarily have the largest budgets.
They will have the clearest understanding of where value is created.
Just as Moneyball challenged conventional baseball wisdom, AI requires companies to challenge conventional technology thinking.
The winners will not be the firms that buy the most AI.
They will be the firms that understand where productivity, capacity, quality, and growth actually come from.
Then they will use AI to amplify those drivers.
Over the next several years, the gap between AI leaders and AI laggards will not be determined by model selection.
Most firms will have access to the same models.
The winners will be the organizations that identify where operational intelligence lives inside their business, capture it, govern it, and make it reusable at scale.
That is the AI equivalent of discovering where runs come from.
Everyone else is still evaluating players.