You Won't Make any More Money

Everyone wants to know when AI agents will show up in the productivity numbers.

Wrong question.

Everyone (i think) tells the electrification story as "new technology takes time." That's the lazy version.

The real lesson that matters: a cheap new resource pays off when you rebuild the operation around it. Buying it gets you almost nothing!

Factories swapped the steam engine for an electric motor and kept everything else: the belts, the shafts, the floor plan, the building itself.

They electrified the old factory. The gains were pathetic.

The big gains came decades later, when they redesigned the factory FOR electricity. A motor on every machine. Single-story layouts. Machines placed where the work flowed, instead of where the drive shaft ran.

If you run a home service company and you read this newsletter, you've already lived this movie once.

It was called global talent.

The swap version: you hired global talent to answer phones for a third of the cost. Same workflows. Same org chart. Same chaos. More affordable.

You saved money. Fine. Take the win.

But that's the electric motor bolted to the old shafting.

The redesign version asks a different question: if full-time, college-educated help costs a fraction of local labor, what becomes affordable that never was before?

Phones answered 24/7, including the 9pm emergency call your competitor's voicemail just ate. A dispatcher who only dispatches. Every quote followed up. Three times. Every completed job gets a review ask. Every invoice chased until it's paid. Every customer called back the day after the job.

None of those roles existed in your company. At local labor prices, they never paid for themselves. So the work just didn't happen.

The constraint moved. The owners who noticed rebuilt the org chart. The owners who didn't got slightly lower cost chaos.

Now AI agents are showing up, and everyone is about to make the same mistake again.

Most companies are about to "agentify" the old office.

Same org chart. Same approvals. Same CSR triaging everything by hand. Same CRM fields nobody trusts. Same owner doing quotes at 9pm. Same 17-step process, except now step 8 is "ask the agent."

Then they'll be shocked when nothing magical happens.

"Where are the productivity gains?"

Buddy, you strapped a motor onto a donkey cart.

The revolution starts when the workflow stops assuming a human has to sit in the middle of every handoff.

A call comes in at 9pm. The agent answers and books the job. Checks the service area and the schedule. Pulls the customer's history. Preps the tech's brief for the morning. Texts the confirmation. Follows up the open quote. Chases the invoice. Asks for the review. Routes the weird stuff to a human.

Your global team stops doing clerical cosplay and handles what actually needs a human: the furious customer, the judgment call, the job that doesn't fit the script.

And you do the owner work: pricing, standards, relationships, the handshake that closes the commercial account.

Agents do the mechanical work. Your global team does the human work. You do the scarce work.

That's the new factory layout.

A few more lessons from the dynamo story:

Replacement is the small prize. Global talent doing your old job description less expensive. An agent doing your talent’s job faster. Both save money. Neither is a new operating model. The early electric motor saved some coal, too. The unit drive changed the factory.

The payoff depends on the complements. Electricity needed new motors, new layouts, and managers who understood what was suddenly possible. Agents need clean enough data. A pricebook that's actually current. Job notes that exist. Someone who can define the work precisely. The system is the system. The agent is one part of it.

Measurement will lie at first. Answered calls. Fewer dropped balls. Quotes that stop dying in someone's inbox. Reviews that actually get asked for. All real money. None of it shows up as a neat line on the P&L on day one.

Standards matter. What can the agent decide alone? What must a human approve? What happens when it's unsure? Who owns the failure? No standards, no scale.

The temptation is to wait for smarter AI to fix all this. It won't. The bottleneck is operational imagination.

Most owners will use global talent and agents to make the current mess lower cost and faster.

The good ones will ask a harder question: "If great help cost a third of what I pay now, and agents were cheap and reliable and always on, how would I design this company from scratch?"

That's where the money is.

You get the productivity revolution by rebuilding the work.

Passage of the week

An engineer at the Crocker-Wheeler Electric Company, writing in 1901, on what factories actually got when they electrified:

He sold them on the coal bill. Swap "coal bill" for "payroll" and it's every global talent pitch you've ever heard. The savings get you in the door. The redesign is the prize.

Yallah Habibi,

Jon

P.S. The factory story comes from economist Paul David's 1989 paper, Computer and Dynamo: The Modern Productivity Paradox in a Not-Too-Distant Mirror. Worth the read if you want the long version.