Everybody Adopted Moneyball. The Edge Lasted Five Years.

THE NUMBER: 10x the individual productivity improvement AI is delivering right now, according to Hebbia CEO George Sivulka. The firm-level productivity improvement? Zero. Same thing happened with electric motors in the 1890s. The gap lasted 30 years.

John Henry, the owner of the Boston Red Sox, told Billy Beane something in 2002 that every CEO in America should hear this week: “Anybody who’s not tearing their team down right now and rebuilding it using your model, they’re dinosaurs. They’ll be sitting on their ass on the sofa in October, watching the Boston Red Sox win the World Series.”

He was right. Within a decade, every team adopted sabermetrics. The Red Sox won in 2004. Then the Cardinals did it. Then the Rays. Then everyone. And the lasting edge from all that statistical revolution? There wasn’t one. The only durable advantage in baseball turned out to be the willingness to pay for star pitching, the one input nobody could quantify.

AI is about to run the same play across all of corporate America, except this time, the transition won’t take a decade. It’s happening in months. George Sivulka (CEO of Hebbia, backed by a16z) published a piece this week that nailed the problem: AI has made individuals 10x more productive. No company has become 10x more valuable. He traced the pattern to the 1890s, when factories bolted electric motors where steam engines used to sit, kept the same floor plan, and wondered why nothing changed. It took 30 years before anyone redesigned the factory around what electricity actually made possible.

Bret Taylor (former co-CEO of Salesforce (NYSE: CRM), chairman of OpenAI) put the same idea in different language: “The atomic unit of productivity in AI is a process, not a person.” What takes 17 days across departments, through legal, finance, IT, procurement, collapses to 17 hours when an agent runs the whole thread simultaneously. You’re not buying a digital employee. You’re buying the ruthless compression of time.

So the imperative is clear. Tear it down. Rebuild AI-first. Anyone who doesn’t is a dinosaur.

But here’s what nobody’s asking: what happened after everyone adopted Moneyball? And what happens to the consumers, the voters, and the communities when every corporation in America runs the same play at the same time?

The answer is Henry Ford’s most important insight, running in reverse.

You Swapped the Motor. You Haven’t Redesigned the Factory.

Sivulka’s piece (amplified by a16z to 52,000 views in hours) traces a pattern that should make every executive uncomfortable. In the 1890s, manufacturers rushed to replace steam engines with electric motors. Same machines, same layout, same workflow. They bolted the new power source onto the old architecture and waited for productivity to soar.

It didn’t. For 30 years.

Gains only arrived when a new generation of owners redesigned everything. Single-story buildings instead of multi-floor (steam engines needed to be on the ground floor; electric motors didn’t). Individual machine drives instead of centralized power shafts. Assembly lines that followed the logic of the product, not the logic of the power source.

That’s where most companies are with AI right now. They’ve swapped the motor. The org chart is the same. The approval chains are the same. The sprint cadence is the same. They’re running electric motors in a steam-engine factory and wondering why their $500K in AI spend hasn’t moved the needle.

Meanwhile, a marketer named J.B. posted this week (130,000 views on X) that he used Claude Code to build a bulk ad launcher for Meta (NASDAQ: META), TikTok, and Snapchat. Configure once, push to all three platforms, live in minutes. What used to take 2+ hours across three different ad managers now takes 10 minutes. He didn’t tweak his workflow. He eliminated three-quarters of it.

And Robert Scoble amplified a piece this week from Aditya Agarwal, one of Facebook’s earliest engineers, who wrote that everything he knows about knowledge work changed in the past two weeks working with Claude. Twenty years of programming experience, and the playbook just got rewritten. Scoble’s frame was blunt: “Either you get new skills and become part of the new world or you will really struggle.”

The plumber we wrote about Monday, the media buyer, and the early Facebook engineer all did what the 1890s factory owners didn’t: they redesigned the work around the tool, not the other way around.

What this means for your business: Stop asking “how do we add AI to our processes?” Start asking “if we built this company from scratch today, with these tools, what would it look like?” If the answer is different from what you have (it is), the question becomes how fast you’re willing to close the gap.

Every Team Adopted Moneyball. The Edge Lasted Five Years.

Here’s the part of the Moneyball story nobody tells. Billy Beane’s Oakland A’s used sabermetrics to compete with the Yankees despite spending one-fifth as much per win ($260,000 vs. $1.4 million). The Red Sox adopted the model and won the World Series in 2004, breaking an 86-year curse. John Henry was right. The dinosaurs went extinct.

But then something happened that doesn’t fit the triumphant narrative. Everyone adopted sabermetrics. Every team hired analysts. Every front office built statistical models. Within five years, the structural advantage disappeared entirely. The only lasting edge in baseball turned out to be the willingness to pay for star pitching. Pure talent. Judgment under pressure. The stuff that defies quantification.

The AI version of this is already becoming visible. Cursor is raising at $50-60 billion. Replit just raised $400 million at $9 billion. Every company on earth will have agentic workflows within three years. When everyone has the same tools, the tools stop being the advantage.

So what’s the AI equivalent of star pitching? It’s the stuff Ethan Mollick keeps pointing to on the jagged frontier: the ability to plan an overall process, iterate within each piece, stress-test the whole. It’s pattern recognition built over decades, not months. It’s knowing which number doesn’t smell right before you can explain why. It’s the experienced operator who looks at an AI-generated deliverable that’s formatted perfectly, cited properly, and completely wrong in ways that take 20 years of domain expertise to catch.

Andrej Karpathy said it this week: “The basic unit of interest is not one file but one agent. It’s still programming.” The skill just moved up a level. The companies that invest in orchestration talent (the people who can conduct the agent orchestra) will have the 2026 version of star pitching. Everyone else will have the same tools and nothing to show for it.

The signal: The moat isn’t AI adoption. It never was. The moat is the judgment that tells the AI what to do, catches it when it’s wrong, and knows the difference between a confident answer and a correct one. Invest in the people who have that. They’re about to become the scarcest resource in your organization.

The $70 Barrel: The Sweet Spot Nobody’s Looking For

Billy Bob Thornton’s character in Landman describes the equilibrium price for oil: around $70 a barrel. At that price, producers get rich and consumers don’t feel pinched at the pump. Go much higher, the economy convulses. Go much lower, the wells shut down. The whole system depends on finding and holding the sweet spot.

There’s an equivalent sweet spot for AI-driven corporate reorganization. Almost nobody’s looking for it.

Henry Ford understood this a century ago. In 1914, he doubled his workers’ wages to $5 a day. Not because he was generous. Because he needed them to buy the cars they were building. The entire consumer economy of the 20th century was built on Ford’s insight: your workers are your customers.

AI runs the Ford equation backward. Every corporation is individually rational to compress headcount, automate hand-offs, and replace the coordination layer with agents. Taylor’s 17-day process becomes 17 hours. Oracle (NYSE: ORCL) cuts 30,000 developers. Amazon (NASDAQ: AMZN) mandates 80% AI usage. The productivity math is irresistible.

But what happens when everyone runs this play at the same time? You haven’t just reorganized your company. You’ve reorganized the economy. College graduates with $200,000 in debt can’t find entry-level jobs because the entry level is being handled by a $20 subscription. Housing is already unaffordable. The first trillion-dollar companies coexist with a generation that can’t afford a starter home.

Americans have always tolerated inequality. They tolerated it when Carnegie built libraries and Rockefeller built universities. They tolerated it because the social contract held: work hard, get educated, do better than your parents. That contract is cracking before AI even starts displacing at scale. AI lands on top of a social contract that’s already fraying.

And here’s the part the Silicon Valley crowd doesn’t want to hear: agents don’t vote. Displaced workers do. Every time. Every historical wave of mass displacement (the Luddites, the Dust Bowl migration, the Rust Belt’s collapse) produced political backlash that reshaped the rules. Not because politicians are noble. Because the incentive structure demands it. An angry, motivated electorate will vote to tax, regulate, and restrict the companies that displaced them.

You can see the early tremors. Seattle’s mayor is trying to prevent grocery stores from closing. New York is layering on employer taxes. These aren’t rational economic policies. They’re the political immune response to economic disruption. And they will get louder, not quieter, as agent deployment scales.

Connect the dots: The companies that treat workforce transition as a line item (two weeks severance, here’s a box) will get regulated into a much more expensive version of the same thing, on terms dictated by politicians who understand AI a lot less than you do. Ford’s $5 day wasn’t charity. It was the smartest business decision of the 20th century. The AI equivalent is figuring out what you owe the workforce you’re reorganizing around. Not because it’s virtuous (though it is). Because the alternative is death by a thousand cuts from legislators and consumers who don’t care about your agent architecture. They care about their mortgage.

What This Means For You

Three forces are colliding, and they point in the same direction. The imperative to reorganize is real (the dinosaurs will go extinct). The advantage from reorganizing is temporary (everyone will do it). And the consequences of doing it carelessly are existential, not for your company alone, but for your market.

Redesign the factory, not just the motor. The 10x individual productivity gain evaporates inside a 2015 org chart. The companies pulling ahead aren’t adding AI to existing workflows. They’re eliminating the workflows entirely.

Invest in star pitching. When every company has agents, the differentiator is judgment, taste, and the ability to catch what AI gets wrong. Those people are your star pitchers. Pay for them. Lose them at your peril.

Find the $70 barrel. There’s a pace and a method of reorganization where the company captures the gain without triggering the immune response. Blow past it, and you’ll spend the next decade in hearings explaining why you prioritized stock price over community.

And here’s the thing nobody’s saying. A friend of ours, a high school dropout, built a complete marketing system on Claude Cowork over 48 hours last week. Costs him $500 a month to run. He’s worth more right now than 90% of the Ivy League graduates in his industry. The tools to rebuild aren’t reserved for the incumbents. They’re available to everyone, everywhere, right now. The same force that’s threatening legacy companies is handing individuals more power than any generation in history has ever had.

It’s always darkest before the dawn. But only if the dawn actually comes for everyone.

Three Questions We Think You Should Be Asking Yourself

If I built this company from scratch today, with the tools available right now, how many people would I hire? If the answer is less than what you have, the math is already working against you. The board has this number in their heads even if they haven’t said it yet. Get ahead of it or get surprised by it.

What’s my transition plan for the people this reorganization displaces? Not two weeks severance. Not a LinkedIn post about how much you value them. An actual plan that accounts for reskilling, redeployment, and community impact. The companies that get this wrong will spend more on regulatory compliance and reputation repair than they would have spent doing it right.

When everyone has agents, what’s my star pitching? The tools will be table stakes within three years. The question that determines 2029 market position isn’t “do we have AI?” It’s “what do we have that AI can’t replicate?” If you don’t have a clear answer, your competitive advantage has an expiration date.

“Anybody who’s not tearing their team down right now and rebuilding it using your model, they’re dinosaurs. They’ll be sitting on their ass on the sofa in October, watching the Boston Red Sox win the World Series.”

— John Henry, owner of the Boston Red Sox (as depicted in Moneyball)

— Harry and Anthony

Everybody Adopted Moneyball. The Edge Lasted Five Years.

THE NUMBER: 10x the individual productivity improvement AI is delivering right now, according to Hebbia CEO George Sivulka. The firm-level productivity improvement? Zero. Same thing happened with electric motors in the 1890s. The gap lasted 30 years.

You Swapped the Motor. You Haven’t Redesigned the Factory.

Every Team Adopted Moneyball. The Edge Lasted Five Years.

The $70 Barrel: The Sweet Spot Nobody’s Looking For

What This Means For You

Three Questions We Think You Should Be Asking Yourself

Sources

Past Briefings

Karpathy Says Stop Coding. A Fastenal Vending Machine Explains Why He’s Right.

Who Checks the Checker? The correction loop is the most valuable thing in AI right now. Nobody is capturing it.

The Plumber Figured Out AI Before the Enterprise Did