The 10X Test

Speed is overrated when it just helps you fail faster.

I really, really want to handle my monthly invoice collection ten times faster. I’m working on it. Claude has been a great help so far, speeding things up by matching the invoices with the bank transactions.

But not everything needs to be done faster. My keynotes, for example, still take between 30 and 60 minutes to deliver. Makes no sense to compress them into five minutes. I can, however, use AI to make those keynotes ten times better.

And when a thing cannot be done ten times faster or better, you can always aim for ten times safer (driving a car, delivering a baby) or ten times cheaper (writing a book, making an animated movie).

Many people think AI saves them time. Often, it just hands them extra bugs and a polished hallucination.

I still like The 10X Test, though.

Most leadership sessions tell you what’s changing. Few tell you what to do about it.

On June 12th, I’m joining Bergman Engineering Masterclass.
Topic: What AI actually means for leaders and teams.
Live, online, 11:00 AM.

Designed for engineering teams — but the first 50 of my followers can join free. Want to buy access for your whole team? Either way, write to team@bergman-masterclass.com.

Your move.

These days, if you cannot make a task, process, or value stream ten times faster, aim for ten times better, ten times safer, or ten times cheaper. The point is your ambition, not the exact arithmetic. Five times, twenty times, a hundred times … who’s counting?

“I saved five minutes” off my daily routine is cute when the thing normally costs me two hours. That’s a 9% improvement.

However, it’s fantastic when before it cost me six minutes. That’s an 83% improvement.

As a solo operator, you live inside the Theory of Constraints. There’s always another bottleneck with your name on it: your time, your attention, your judgment, your stamina on a bad day after three client calls and one long tax form. Technology matters when it hits that constraint hard enough to change the entire economics of your week.

That’s why I care little for the AI-induced nonsense on LinkedIn. “Look, I made an infographic with nine pictograms, eighteen arrows, and twenty-seven labels in only five seconds.” Lovely. You also made the internet slightly worse.

The more serious question is: where does AI create real change, and where does it only make you faster on the way to dumb and dumber?

Fortunately, the available data already gives us a rough map.

In multiple studies, people using generative AI on various tasks finish faster and produce work that’s rated higher. But we’re talking single or lower double-digit improvements. That’s good. It’s also a cold shower for the 10X hype merchants, because “10X” rarely happens overnight.

Most improvement gains from AI come as compounding bumps, not fireworks. You get 1.4x here, 1.8x there, and after a year of relentless struggle, your shabby little bicycle turns into an enviable scooter.

That still matters. For one-person businesses, a steady lift in output and quality can mean one extra client, fewer revision loops, less rework, and more sleep. I’d gladly take that.

But speed has a dark little cousin: false confidence.

In other studies (that I cannot be bothered to link to here) researchers found that developers using AI assistants wrote less secure code. That’s the nasty part. The machine gives people faster mistakes with a confident smile.

Which is why going 10X faster is not always a goal that makes sense.

Owning the fastest car on the planet is rather pointless when local law still imposes a 30 km/hour speed limit on your street.

Sometimes, your time and attention are better served trying to do things ten times better, ten times safer, or ten times cheaper. Not ten times faster.

Jurgen, Solo Chief.

P.S. What are you doing ten times faster, better, cheaper, or safer?