This week I sat down to write a memo for my CBL colleagues, and halfway through I realized I was really writing about the thing I've been circling for two decades of project management: there are two completely different jobs hiding inside the job title "project manager," and I've spent most of my career treating them as one.
In Japanese I split them as koto and hito. Koto — literally "the matter" or "the thing" — is the schedule, the budget, the risk register, the status report, the process. Hito — "the person" — is why a stakeholder is stalling, why a team goes quiet in meetings, why two departments that should cooperate instead quietly sabotage each other. For most of my career the two blurred together into one skill called "project management," and the koto side was where the credentials, the Gantt charts, and the professional identity lived. Hito was the messy overflow you handled with instinct, if you handled it at all.
What made me actually write the memo, rather than just think it, was watching how fast AI has started eating the koto side this year. Schedule optimization, cost forecasting, risk analysis pulled from historical data, progress reports auto-summarized from meeting transcripts — the tools I trained a generation of junior PMs to painstakingly build by hand are now something you generate in a prompt. And PMI saw this coming before I did: PMBOK's sixth edition, the one I studied for my own certification, was almost entirely a koto document — process, tools, techniques, prescribed in detail. The seventh edition, published in 2021, pivoted hard toward what they call "power skills" — principles for how a team actually creates value, not a checklist for how to run one.
That's not a coincidence. It's the largest project management body in the world quietly admitting that the technical half of the job was becoming commoditized well before AI made it obvious.
Here's the part that actually matters, though, and it's the part I underestimated for years: AI is extraordinarily good at statistical patterns, and hito problems are not pattern problems. "Why won't this team move" isn't sitting in a dataset — it's sitting in a history between specific people, in resentments nobody wrote down, in a culture gap nobody named out loud. You cannot prompt your way to that diagnosis. You have to be in the room, and you have to actually know the people.
This week an AI tool drafted a full project schedule, a cost forecast, and a risk register in a fraction of the time it used to take — work that once ate a full day of mine, back when I built these things by hand. I sat there a little stunned, and a thought I've circled for two decades finally clicked into words: project management has always been two different jobs stitched together, and I'm not sure the one I trained for is the one that matters anymore.
In Japanese I split them as koto and hito. Koto — literally "the matter" or "the thing" — is the schedule, the budget, the risk register, the status report, the process. Hito — "the person" — is why a stakeholder is stalling, why a team goes quiet in meetings, why two departments that should cooperate instead quietly sabotage each other. For most of my career the two blurred together into one skill called "project management," and koto was where the credentials, the Gantt charts, and the professional identity lived. Hito was the messy overflow you handled with instinct, if you handled it at all.
I'm not the only one who's noticed the split widening. PMBOK, the project management bible I studied for my own certification, was almost entirely a koto document in its sixth edition — process, tools, techniques, prescribed in detail. The seventh edition, published in 2021, pivoted hard toward what PMI calls "power skills" — principles for how a team actually creates value, not a checklist for how to run one. The largest project management body in the world quietly admitted the technical half of the job was becoming commoditized well before AI made it obvious.
Here's the part I underestimated for years, though: AI is extraordinarily good at statistical patterns, and hito problems are not pattern problems. "Why won't this team move" isn't sitting in a dataset — it's sitting in a history between specific people, in resentments nobody wrote down, in a culture gap nobody named out loud. You cannot prompt your way to that diagnosis. You have to be in the room, and you have to actually know the people.
So the koto knowledge I spent years accumulating — the tools, the methods, the certifications — is depreciating in market value in real time, whether I like it or not. But the hito work — reading what actually motivates a stakeholder, building a room where people function as themselves instead of performing a role, turning a cultural gap into leverage instead of friction — isn't just AI-proof. Its scarcity is going up precisely because everything around it is getting automated.
The broader takeaway, for me, is uncomfortable and clarifying at the same time: the parts of this career I was proudest of on my resume are the parts getting automated first, and the parts I used to file under "soft skills" are turning out to be the whole point. If that schedule is any indication, the koto side isn't coming back as a differentiator. The only question left is how deliberately I invest in the hito side instead of treating it as something that just happens if I'm lucky.





