Chairman and CEO of Microsoft
A Line That Stopped Me
Microsoft CEO Satya Nadella published something this week that stopped me mid-scroll.
Not because it was surprising.
Because it gave me a new way to understand a writer who lived nearly seven hundred years ago.
https://x.com/satyanadella/status/2066182223213293753?s=46
Nadella’s argument, stated simply, is that in an AI-driven economy, tasks can be outsourced. Entire jobs may eventually be outsourced.
Learning cannot.
In his words:
“You can offload a task, or even a job, but you can never offload your learning.”
The line stayed with me.
Not because of what it says about AI.
Because of what it says about people.
And because, unexpectedly, it reminded me of Yoshida Kenkō.
A Monk With No Audience
Kenkō was a Japanese monk and essayist who lived in the fourteenth century.
Today he is remembered for Tsurezuregusa, usually translated into English as Essays in Idleness.
The title sounds leisurely. The book is anything but.
It consists of 243 short fragments—observations, reflections, anecdotes, complaints, philosophical musings—written over many years.
There is no grand system.
No argument being constructed.
No audience strategy.
No publishing plan.
Just a man observing the world and writing down what he noticed.
By modern standards, it looks remarkably inefficient.
Yet seven centuries later, people are still reading him.
What Survives in Essays in Idleness
When I first encountered Tsurezuregusa, I thought it was a book about impermanence.
And it is.
But reading Nadella’s essay made me realize that it is also a book about something else.
Accumulated judgment.
Every fragment begins with an observation.
Something seen.
Something remembered.
Something that felt worth noticing.
But the observation is never the point.
The writing is.
The act of turning experience into language forces Kenkō to decide what he thinks about it.
And because he repeats that process hundreds of times, across years and changing circumstances, the fragments begin to form a pattern.
Not a theory.
A mind.
What survives in Tsurezuregusa is not information.
It is judgment.
The Organizational Version of the Same Problem
That, I think, is where Nadella and Kenkō unexpectedly meet.
Nadella is describing the organizational version of a very old practice.
Organizations learn through repetition, reflection, and memory.
Or at least they try to.
The problem is that most knowledge remains trapped inside individuals.
A veteran salesperson retires.
A great teacher leaves.
An experienced manager moves on.
The knowledge goes with them.
What AI makes possible, Nadella argues, is the creation of systems that capture and compound that learning rather than letting it disappear.
He calls this “token capital.”
Not merely access to an AI model.
A learning system built on top of human experience.
The model can change.
The learning remains.
The moat is not the model.
The moat is the learning history.
Writing as a Learning Loop
The more I thought about it, the more I realized that writing serves a similar function for individuals.
I have spent years writing essays on note, a Japanese publishing platform.
Most of them reach relatively few readers.
That has never bothered me very much.
Because their primary value was never distribution.
It was accumulation.
Each essay forces me to convert something I have read, taught, experienced, or noticed into language.
Sometimes I discover that I understand the subject less than I thought.
Sometimes I discover a connection I could not see before writing.
Sometimes I change my mind entirely.
The essay becomes a record of that movement.
Over time, those records accumulate.
Not as information.
As judgment.
The Difference Between Consuming and Converting
This is why I suspect the most important part of Nadella’s argument is not actually about AI.
It is about conversion.
Most people consume knowledge. Far fewer convert it.
Reading feels productive.
Watching videos feels productive.
Collecting notes feels productive.
Sometimes it is. Often it is not.
Knowledge only becomes durable when it is forced through some kind of friction.
Teaching it.
Defending it.
Applying it.
Writing it down.
That friction is where judgment forms.
And judgment is what compounds.
What Actually Compounds
The technology has changed dramatically since Kenkō’s time.
The underlying principle has not.
A company can outsource a task.
One day, perhaps, it can outsource an entire job.
But it cannot outsource the learning that comes from doing the work, reflecting on it, and building systems that preserve what was learned.
The same is true for individuals.
AI can help us write.
It can help us research.
It can help us think.
But it cannot accumulate our judgment for us.
That still requires attention.
Reflection.
Experience.
And the willingness to return to the page.
Seven hundred years ago, a monk sitting in a hut outside Kyoto understood this without using any of the language we use today.
He simply kept writing.
The loop kept running.
That may be why he is still being read.
And why Satya Nadella’s observation feels less like a prediction about the future than a reminder of something very old.
You cannot outsource learning. ◾️
First published on Substack on Jun 16, 2026 (JST). Archived here as part of the Fabio Caipirinha permanent archive (published on this site: Oct 15, 2026). Images from the original are omitted here pending rights clearance. Original: https://fabiocaipirinha.substack.com/p/you-can-outsource-the-job-but-you
Cover image: Okumura Masanobu, “A Stylish Tsurezuregusa” (late 1710s) (Public domain, Wikimedia Commons — source)
