For about forty years, electricity did almost nothing for the American factory.
The power was there. Edison opened his first station in Manhattan in 1882, and within a decade a factory owner could buy all the electricity he wanted. Most of them bought it and changed nothing else. Out came the steam engine, in went one big electric motor, driving the same overhead shaft, the same belts, the same machines bolted in the same rows. The energy bill dropped a little. Nothing else did.
In 1900, electric motors turned less than five percent of the machinery in American factories.
The jump came when somebody stopped swapping the engine and started redrawing the floor. Put a small motor on each machine and the shaft is unnecessary. Without the shaft you can arrange machines in the order the work actually flows, build low and wide instead of tall, cut windows into the roof, and run a crane over the top. Output per worker climbed through the 1920s. By 1930, motors ran more than eighty percent of factory machinery.
The technology arrived in 1882. The idea of what to do with it arrived around 1920.
We’re in 1882.
Everyone I talk to agrees AI changes everything. Almost nobody can tell me what they would actually build with it. That gap is where the money is, and it isn’t new. It opens at the start of every technology era, and it closes fast.
Eras get won by whoever works out what the new thing is for. That’s rarely the person who built it.
- Amazon: e-commerce already existed when Bezos started. He made buying from a website you’d never heard of feel safe, then made it cheaper and faster than driving to a store.
- Google: search engines already existed. Google made the results relevant, then put an advertiser at the end of every question a person typed.
- Netflix: video rental already existed. Netflix killed the late fee, then killed the trip, and turned a per-movie transaction into a monthly relationship.
- Uber: paying a stranger for a ride already existed. Uber made it instant, priced up front, and reliable enough that you stopped keeping a taxi number in your phone.
- Airbnb: renting a room from a stranger already existed. Reviews and a guarantee made two strangers trust each other, and an empty bedroom became income.
- Stripe: payment processing already existed. Stripe made it seven lines of code, which turned every developer into someone who could take money.
- Instagram: photo sharing already existed. Filters made everyone’s photos good enough to post, and a private moment became a public performance.
Every one of them built a new economic relationship inside a category that was already there. The technology was the easy half.
Two strengths, and most people have one
Recognizing an important new idea is the first. Commercializing it to the masses is the second.
They’re separate skills, and they rarely live in the same person. That’s why whoever sees it first so seldom ends up owning it.
Being early is a hobby until someone can buy the thing.
What “better” actually means
An important idea changes how we do something in a significantly better way. Better comes from three places:
- Speed: what took a week takes an hour.
- Price: what cost 500€ costs 5€.
- Experience: what was miserable becomes pleasant.
Move one of those hard, or nobody has to switch. People don’t switch out of admiration. They switch when staying costs them more than moving.
Getting it to the masses is then two more jobs, and skipping either one kills the company.
Make it usable. It has to work for a person who doesn’t care how it works, work every time, and cost little enough that trying it isn’t a decision.
Distribute it. Creatively, and often enough that people change what they already do. Existing habits are the real competitor.
I spend most of my time on the first job. Building mailover, I’ve had weeks where the product got meaningfully better and exactly nobody new found out, and I still caught myself calling that a good week. Half the battle is building the thing. The other half is letting the world know it exists, and that half doesn’t get easier by ignoring it.
The friction is the business
Here’s the pattern underneath all seven of those companies.
Somewhere there’s an exchange both sides want. A person wants the thing, someone can supply it, and both would be better off afterwards. It doesn’t happen, because something sits in the middle: cost, time, distance, distrust, or the sheer hassle of arranging it.
Winners find a valuable exchange that should be happening and isn’t, then remove whatever is stopping it.
Before Airbnb, millions of people had a spare room and millions wanted somewhere cheap to sleep. The exchange didn’t happen because neither side would trust a stranger with a key or a credit card. Reviews and a guarantee removed exactly that, and an empty bedroom turned into an asset.
Nothing about the bedroom changed. The relationship around it did.
Every era has an input that goes to nearly free
Zoom out and the same story repeats one level up. An era starts when some input becomes abundant and cheap, and it goes to whoever works out what the input is for.
Rockefeller and hydrocarbons. Oil got cheap and Standard Oil became the default energy of the industrial world. Look around the room you’re in: most of the objects in it are made of, moved by, or wrapped in hydrocarbons.
General Electric and electricity. Power got cheap and GE sold households appliances. The washing machine handed back hours of every single day to the women doing that work by hand. Those hours went into paid work, into study, into a life with more choices in it. The appliance changed the house before the grid changed the economy.
Steve Jobs and silicon. Chips got cheap and Apple put a computer on a desk in an ordinary home instead of in a lab. Now most of us carry a second one in a pocket.
Jeff Bezos and shelf space. The internet made a storefront and a distribution channel cost almost nothing. Amazon worked out what to do with that, and now nearly everything is sold there.
The next line of that list has a blank in it.
Intelligence is going to nearly free. Expert judgment, the thing that cost 300€ an hour and required an appointment, is collapsing toward the price of electricity. Somebody is going to find the appliances.
The light bulb is not the appliance
Electricity’s first product was the light bulb. It replaced the candle, one for one, doing the same job better. Useful, obvious, and not what won the era.
The washing machine won the era, because it had no candle to replace. It did something no household could do before at that price.
Right now the chatbot is our light bulb. It sits where the search box and the blank document used to sit, and it does that job better. Every model company sells one and they all look the same, which is the tell.
The appliances haven’t been built yet.
What isn’t happening today
Which valuable exchanges aren’t happening right now, and would start happening if judgment were free?
Five kinds of friction are about to get cheap. Each one is a category.
- Expert judgment costs too much. A landlord with one flat and a fourteen-page contract doesn’t call a lawyer for a twenty-minute question. The question goes unasked, and a bad clause gets signed. Multiply that by every small decision every small business makes alone.
- Processing takes too long. A clinic has eight years of patient notes it will never read again, because reading them costs more than the answers are worth. The value is sitting there, priced out by time.
- The expertise isn’t where the person is. A farmer with a sick crop, a founder in a country with three good lawyers, a parent with a kid who learns differently. The expert exists, just not within reach.
- People can’t say what they need. Search demands you already know the word for the thing. Most people with a problem don’t have the word. Everything that has ever been sold through a knowledgeable shopkeeper sits in this category, and the shopkeeper doesn’t scale.
- Trust hasn’t been established. Two parties who would happily transact and have no way to verify each other. Airbnb solved one instance of this with reviews. Hundreds are still open.
Pick one, in one industry, and you have something to build on Monday.
A thousand phds in your pocket
That’s the question I come back to when I’m stuck on what to build.
Picture a thousand tireless experts, all awake at once, all working on a job you understand and most people don’t. What do you have them do?
Any answer that starts with “a chatbot that” is still the light bulb.
Two more questions, in order. What does your ideal future look like? Describe the version of your work or your day that you actually want, then go and build that instead of waiting for someone else to.
And which experience you use every week could be 10x faster, cheaper or better? Weekly matters. If you don’t feel the pain often, you’ll quit before you solve it.
Where to start
Pick a niche where you have an unfair advantage. Knowledge, relationships, or a genuine obsession. In a market moving this fast, the durable edge is knowing something the crowd still has to go and learn.
Build something people can use today. A grand vision for 2030 is a way of never being wrong. Ship the small thing that works this week, and let real usage tell you what the big thing is.
Get to real customers fast. The market moves faster than any roadmap you write. Long development cycles are how you arrive with a perfect answer to last year’s question.
Find the exchange that should be happening. Work out what’s blocking it. Remove that, for a specific group of people, before anyone else does.
You already have a thousand experts in your pocket. Most people ask them to write emails.
What would you ask?