The Future of Work Was Written in 1645

Everyone is asking what AI will do to work. Wrong question. The better one: when machines can do the doing, what is left that is actually yours?

I have spent 25 years delivering technology across more than 20 countries. I have watched entire skill sets go from career-defining to irrelevant. Mainframe operators. Server administrators. Ticket dispatchers. Every wave produced the same two groups: people who had built their identity on a tool, and people who had built it on judgment. The first group panicked. The second group adapted and moved up. AI is running that same movie again, just faster and for everyone at once, including the knowledge workers who thought they were safe.

Which is why the best future-of-work guide I have ever read is not a Gartner report or a McKinsey deck. It is 380 years old, written by a swordsman on one sheet of paper.

One sheet of paper, 21 lines

In the last weeks of his life, Miyamoto Musashi sat down and wrote 21 short precepts. By his own account he had fought more than sixty duels and was never beaten. He had lived through the collapse of an era: the warring states period ending, the samurai class losing its function, an entire profession watching its core skill lose market value in a single generation. Sound familiar?

He called the text the Dokkodo, “The Way I Go Alone.” It was not a sword manual. He had already written that one. This was something else: a personal operating system, the rules he had run himself on for decades, compressed to their minimum. He handed it to a single student and died.

Three hundred and eighty years later it reads like the only honest guide to the AI age I have found. Not because Musashi predicted machines. Because he understood the one thing the future-of-work debate keeps missing: tools change, the operator’s rules do not. AI is changing what work is. It is not changing what makes a person worth hiring, following, or trusting. That part has not moved in centuries. Musashi wrote it down. Here are six of his lines.

“I will give preference to nothing among all things.”

Tech-agnostic before it was a LinkedIn buzzword. The people who tied their identity to the last cycle’s hot tool are the ones panicking now. I have seen this on delivery floors for two decades: the engineer who is “the ServiceNow person” or “the UiPath person” versus the engineer who is the person who solves the problem with whatever is on the table. The second one survives every platform migration. The first one fights every migration, because the migration threatens who they are.

Do not fall in love with a model, a vendor, or a stack. The model you are loyal to today will be outperformed within a year, and switching costs are collapsing. Love the outcome. The tool is disposable. Your judgment is not.

“I will not own anything that will one day be a valuable antique.”

The most uncomfortable line for any expert to read right now. The skill you spent a decade mastering may be next year’s antique. Musashi refused to collect things that only gain worth sitting on a shelf.

Apply it to your CV. Most professionals carry their skills the way collectors carry antiques: proud of the acquisition cost, blind to the current market price. The decade you invested is a sunk cost. The market does not pay for what a skill cost you. It pays for what the skill produces today. Audit yourself once a year, honestly. Keep the skills you use. Drop the rest without grief. Grief is expensive, and the market does not refund it.

“I will not regret my deeds.”

AI-made iteration almost free. The bottleneck is no longer producing the work. It is the hours people burn agonizing over the last version, the pilot that failed, the automation that got rolled back. I have run programs where the post-mortem lasted longer than the project. That math never made sense, and now it is indefensible.

When the next attempt costs nothing, regret is the most expensive habit you can keep. Ship. Learn. Move. Treat every failed output as data, not identity. The teams winning with AI right now are not the smartest ones. They are the ones with the shortest recovery time between attempt and next attempt.

“Buddhas and Gods are worthy of adoration but I will ask them for nothing.”

Read that twice. Musashi revered the gods and refused to beg them for help, because the moment you ask others to win your fight, you lose the thing that made you.

Now swap “gods” for “the model.” Use AI relentlessly. I do, every day, across content, delivery, and operations. But there is a line, and most people cross it without noticing: the moment you stop checking the output because the machine is usually right, you have not automated the task. You have automated yourself out of the decision. Respect the model. Use the model. Never outsource your judgment to it. The person who lets the machine decide has already lost the only contest that mattered. This is the whole game.

“Even if I sacrifice my life I will not sacrifice my name.”

When content is infinite and free, the only scarce asset is the name on it. Anyone can now generate a plausible article, a plausible proposal, a plausible strategy. Plausible is worthless. The market price of plausible just went to zero.

What did not go to zero is the reputation for being right. When a client signs off on my recommendation, they are not buying the document. They are buying the fact that my name is on it and I have to live with the result. AI can generate the words. It cannot generate the accountability behind them. Guard your name like Musashi guarded his. It is the last professional asset that does not depreciate.

“I will never deviate from the way of Heihô.”

Heihô is the way of strategy, his craft. The way has no end. You never arrive. His translator reads the line this way: the value of a human being lies in always remaining open to growth, and never deciding you are perfect and stopping on the way. That is the entire job description for the next decade. Not “learn AI” as a box to tick, one certification and done. The half-life of that certification is measured in months.

Stay on the way. Make learning a standing process, not a project. The professionals who treat AI literacy the way athletes treat training, daily and permanent, will compound past the ones who treat it as a one-time upgrade.

The lines he did not need to explain

The other precepts hold up just as well. He wrote that he would accept things as they are rather than as he wished them to be: the single hardest discipline in any transformation program, where most failures start with leadership refusing to see the real baseline. He wrote against jealousy and resentment: watching a competitor adopt AI faster than you is only useful if it changes what you do tomorrow. And he wrote, at the end of a life spent around death, that he would not fear it. Translate that to a career: the willingness to let a version of yourself end so the next one can start. Most people protect the old version long past its expiry date. That protection is what actually kills careers.

The point

The future-of-work debate keeps staring at the machine. Musashi would tell you to look at the operator. Tools change. Eras change. Swords became rifles became models. What he protected was the human layer underneath: discipline, ownership, judgment, the refusal to cling and the refusal to beg.

AI will take the tasks. Good. Let it. Your work was never the tasks.

I frame this for clients as Output, Outcome, Impact. Output is the task: the ticket closed, the draft written, the code shipped. That is the layer AI is eating, and it should. Outcome is what the output changes: the incident that never recurs, the client that renews, the decision that gets made. Impact is what compounds over years: the reputation, the trust, the name. Musashi’s 21 lines all defend the top two layers. He watched an entire class of professionals discover that their output had lost its market. The ones who survived were the ones whose value was never the output in the first place.

AI took the output layer. It cannot take the outcome. It will never own the impact. That was true when the ground was feudal Japan losing its warrior class. It is true now.

So stop asking what AI will do to your job. Ask what you have built that does not depend on the tasks. If the honest answer is “not much,” you now know exactly what to work on, and no model can do it for you.

Ship the AI. Deliver the outcome. Keep control of the only thing that was ever yours.

A dying swordsman worked this out with a brush and one sheet of paper. We have every tool he never dreamed of. The least we can do is keep the operating system.

Precepts quoted from Teruo Machida’s English translation of the Dokkôdô, Bulletin of Nippon Sport Science University, 41(2), 2012.