Bite into a Kit Kat, and you’re eating failure. Between each perfect wafer is a layer of ground up rejects. That’s exactly how we should build agents.

A market beats a committee. And yet that isn’t how we run our businesses.

Jane Street is a quant trading firm that made $13M per employee last year. They're famously strange. They built a realtime auction to allocate GPU compute using a currency called “Hive Bucks." Anyone can bid, outbid, or kill a project they think is wasting cycles. No approval. No meeting. The market clears itself.

Most companies don't have a compute problem. Yet. But they do have constraints. And the most common solution is a State Planning Commission with better catering.

Galileo hid his greatest findings in a secret code, and it held back science for generations. Now, Gary in Accounting is doing the same thing.

When playgrounds install softer surfaces, kids break MORE arms. Kids see rubber and think they're safe. So they climb higher and jump from places they wouldn't have before.

But broken arms aren't broken necks. Good playground design isn't about eliminating injury, it's about padding the things that cause broken necks, and leaving enough real risk so that kids stay and learn to climb.

If we pad the floor, risk doesn't disappear.
It relocates.
Same thing happens with AI governance.

In 1956, the FBI foiled the biggest bank robbery in US history — just five days before the robbers would have gotten off scot free. They survived six years of the FBI, but they didn't survive each other. 

The gang's rules were simple:
Don't touch the money, don't get arrested, don't talk.
Ride out the six-year statute of limitations and walk free.

They almost made it. What broke them wasn't a detective. It was the split.

I see the same squabble today over who gets credit for AI gains. These fights don’t just poison one initiative, they kill the transformation that’s supposed to follow. 

In 1986, scientists worried runaway nanotech might convert the entire earth into Gray Goo. And that’s exactly what it looks like inside some of my clients.

Phrase of the week: "AI sprawl."
When everyone can vibe code an app, each solution becomes more goo, slowing down the company. How long before tech and org debt eats the business?

Brex may have just open-sourced an antidote.

UnitedHealth Group ran an AI with a 90% error rate. On purpose.

That's what the class action says. The nH Predict algorithm, built on 6M patient records, would automatically override physician recommendations. If patients appealed, they won 9 times out of 10. The courts are now forcing UnitedHealth to answer: Did they build governance around a model they knew was wrong? The answer, from what's emerged in discovery, appears to be No.

That's what ungoverned AI looks like.

Anthropic’s Economic Index just measured millions of Claude chats and published what people actually use AI for at work. Their finding: 49% of U.S. occupations now have at least a quarter of their tasks being handled by AI.We did the math on what those tasks cost…

The inventor of the printing press died penniless.

Not because the press didn't work. It was spectacular. Gutenberg produced in days what took scribes months. He solved the hardest part of the problem…and never saw it coming when his solution became the New Problem.

He flooded his local market with cheap books. Prices collapsed. His books had no way to compete with all the expensive books in faraway towns. He'd revolutionized production but ignored distribution. His breakthrough created the crisis.

Three teams that threw away their old systems to work AI-first.

Not AI-assisted. Not "we gave everyone Copilot." Threw away the process and started over.

There's a surgery called an osteotomy. When a bone has healed wrong, it's functional enough to walk on but misaligned the only fix is to re-break it, set it correctly, and let it heal from scratch. You can't correct the alignment from the outside.

Most workflows are that bone. They healed around fax machines, email chains, weekly standups, and quarterly planning cycles. Layering AI on top doesn't fix the alignment. It just makes the limp a little faster.

In 1923, Henry Ford owned 75% the American tractor market. By 1928—5 years later—he walked away a failure. 

Problem was, Ford’s tractor was just a mechanized plow. He simply replaced the horse with an engine.

John Deere saw something Ford missed:…

In 1921, Billy Mitchell sank an ‘unsinkable’ battleship with an airplane. The Navy's response? Courtmartial him, and promote the admirals who said he couldn’t do it.

"A committee where 10 people have to say yes is the opposite of innovation." That's Coinbase CEO Brian Armstrong. Last year, one innovation bet netted them $1.35B in revenue.

Here's the system that made it possible—and why most companies could never replicate it.

Is AI making teams more robotic? Human-AI teams are 50% more productive. But, they’re about 25% less social. Output’s up, vibes are down.

New MIT/Johns Hopkins study: 2,234 workers. 11,000+ ads created. Ads tested with 5 million impressions on X.

Researchers found something most AI coverage skips…

The Pony Express ran for 18 months.
It shut down two days after the transcontinental telegraph was completed.

The riders knew it was coming—telegraph poles were going up the whole time they were riding. But they rode anyway. Because a letter that arrives in 10 days beats one that never arrives.

In 1961, an 11-year-old girl spent four days floating on a cork raft in the open Atlantic. No food, no water, no shelter from the sun.

The crew that finally found her said the hardest part wasn't the search area—it was spotting the raft. Neutral-colored equipment blended into the waves. A human being, drifting in millions of square miles of ocean, was nearly invisible.

That rescue changed everything.

You don't need an AI strategy. You need wins. I spent 13 years at IDEO watching orgs make the same mistake with design thinking that they're making with AI right now:

They interview stakeholders.
They build a beautiful matrix of prioritized use cases.
Then they pilot whichever ones scored highest on the 2x2.
Six months and a few hundred thousand dollars later, they kill it.

It takes zero imagination to eliminate a job.

It takes tremendous imagination to invent new ones.

Too many companies take the zero approach.

One person plus AI equals a two-person team. Harvard proved it with 776 employees at P&G. I've started calling what happens next "the bozos effect." I see it with almost every client now. Give someone AI tools and within a few weeks they're quietly wondering why they need the rest of the team.

It's not imaginary. A Harvard/GitHub study found developers with Copilot shifted toward solo coding and away from collaboration. Not because anyone told them to, but because they could. 

An AI looked at 96,000 LinkedIn photos. It predicted career success better than GPA, GMAT, or attractiveness. Combined.

New National Bureau of Economic Research study from The Wharton School and Yale University used AI to extract Big Five personality traits from MBA grads' headshots. Then tracked their careers.