You lived the first two. If you were running anything before about 2005 you kept your books in a ledger or a shoebox of dockets, and then a bloke came round selling MYOB and told you the computer would do it for you. Half your mates said it was a fad. The ones who put the till data into a spreadsheet and actually read it started pricing sharper than everyone else, and the shoebox crowd never worked out why they were always a step behind.
Then the internet did it again. Same shape of jump, same fear. The businesses that got a website and took bookings online in 2008 looked like show-offs, and by 2015 the ones who hadnt were invisible.
Both times the pattern held. Scary and disruptive, everyone reckoned it would blow over, and the early movers quietly pulled ahead while the holdouts lost ground so slowly they didnt notice till it was too late to catch up.
This is the third one. Digital to generative. It will feel exactly the same in your gut, which is how you know its real.
David Autor, an MIT economist, put out a paper in the Quarterly Journal of Economics in 2024 on exactly this. He and his co-authors tracked eighty years of job titles and found that about 60 percent of the work people do today sits in roles that didn't exist in 1940. Automation kills off some tasks, sure. It also invents whole categories of work that weren't there before, and the people who move early end up owning those new roles. The bookkeeper who learned spreadsheets in 1999 became the one everyone came to.
Zoom right out and the numbers are almost stupid. For most of human history output per person barely moved, a compound growth rate of about a tenth of a percent a year, near enough to a flat line for centuries. Then the industrial revolution hit and it climbed toward one percent a year, enough to more than double what a single worker produced inside a hundred years. And the gains landed unevenly. They went to whoever moved on it.
We're standing at the front of another one of those lines. Erik Brynjolfsson, who runs the Digital Economy Lab at Stanford, has been measuring this wave while it happens. His customer support study clocked agents resolving 14 percent more per hour, the least experienced ones a third more. He reckons US productivity ran about 2.7 percent in 2025, near double the ten-year average. It's already showing up in the data while people are still arguing about whether it's a fad.
So here's what i'd actually do, owner to owner. Don't panic, and don't go buy forty subscriptions because a podcast told you to. Move the way the good operators moved with computers and the web, early and deliberate. Pick one thing that eats your week, the quoting, the follow-up emails, the invoice chasing, the same three questions customers ask you every day, and get the machine doing the first draft of it. Watch it for a fortnight. Keep the bits that work, bin the bits that dont. Then do the next thing.
The reason to start now instead of next year is the same reason the spreadsheet crowd won. The advantage compounds. The bloke who started reading his own numbers in 1999 had six years of practice by the time it was obvious to everyone. You can't buy that back later. You either put your reps in early or you spend the back half of the decade chasing someone who did.
You already know how this story goes, you watched it twice. The only question is which side you're on this time.