I don't find my work interesting anymore.
That's what a mentee said to me last week.
He's a project manager — also wears the scrum master and product owner hats, depending on the day. Sharp, experienced, the kind of person you want running a delivery team. I asked him a routine question: how are things going, how are you feeling about the work?
He paused before answering. Then: "The work doesn't really interest me anymore. AI is picking up more and more of my tasks. I feel like a scheduler and an orchestrator. Not a contributor."
That response stopped me in my tracks, and I wanted to explore it in more depth.
What's actually going on
What he described deserves a real answer, not a productivity tip. So I want to start with the diagnosis before getting anywhere near a prescription.
A lot of professional identity is built on visible artifacts. The standup you ran. The backlog you groomed. The status report you wrote on Friday. The risk register you maintained. These are the things you can point to and say I did that. They're how you measure a week.
When AI starts producing those artifacts — drafting the status report, generating the burndown summary, suggesting backlog priorities — your week stops feeling like your week. The artifacts are still being made. Just not by you. And even if you know intellectually that the underlying value of your role (judgment, customer context, the calls in the gray area) is still there, the wobble in identity is real.
So the feeling isn't a productivity problem. It's a meaning problem. Worth treating as such.
Knowledge work has layers
Here's a way to picture what's happening.
Think of knowledge work as a stack. At the bottom is execution — the ceremonies you run, the tickets you update, the documents you draft, the meetings you schedule. Above that is coordination — keeping work moving, surfacing blockers, communicating status across teams. Above that is judgment — calls about what matters, where the requirement is wrong, what to prioritize when everything looks urgent. At the top is domain — the deep understanding of the customer, the mission, the why behind the work.
AI is eating this stack from the bottom up. Execution first. Coordination next. It's already encroaching on judgment in narrow, well-bounded cases.
The move — and I think it's the only sustainable one — is to elevate. Move up the stack faster than AI moves up under you. The people who feel like schedulers right now are people who got pinned at a layer that's shrinking. The people who feel energized are people who used the new tools as a lift, not a floor.
That elevation is what the rest of this post is about.
The typesetter's choice
The closest historical parallel I can think of is what happened to typesetters in the mid-1980s.
Before desktop publishing, typesetting was a real craft. A typesetter knew kerning, leading, font selection, hot metal and phototypesetting machinery. They had hands-on expertise that took years to build. When you wanted a printed page that looked professional, you hired one.
Then Aldus PageMaker shipped in 1985. A Mac and a LaserWriter could now produce typeset pages. The artifact — laid-out, professional-looking text on a page — was suddenly available to anyone with a few thousand dollars of hardware and an afternoon to learn the software.
The typesetters who struggled were the ones who defined themselves by the operation of the machinery. I am the person who runs the Linotype. When the Linotype became unnecessary, so did they.
The typesetters who thrived redefined themselves around what they actually understood: how letterforms carry voice, what makes a page readable, why one layout works and another fights the reader. They became designers — and over the next decade, design became more visible and better-paid than typesetting had ever been.
The craft didn't die. The artifact-level skill did. The judgment-level skill became more valuable, not less. In the language of the stack: the surviving typesetters moved up — out of operation, into judgment about how communication actually works on a page.
I think that's the frame for what's happening to PMs, scrum masters, product owners, and a lot of other knowledge workers right now. If you define your job as the person who runs the ceremony and maintains the board, you're a typesetter. If you define your job as the person who makes sure the right thing gets built for the right customer at the right time, you have several directions you can grow into.
What elevation looks like in practice
Concrete is better than abstract here. What does moving up the stack actually look like for delivery roles? A few examples I keep coming back to:
The PM who used to spend Monday morning building the weekly status deck now spends Monday morning sitting with the contracting officer's technical representative — understanding what's actually keeping them up at night, what pressure they're under, what success looks like for them this quarter. The deck still gets built. AI does the first pass. The PM edits and ships it in twenty minutes.
The scrum master who used to chase ticket updates now spends that time pressure-testing whether the team's interpretation of the requirement matches the agency's actual mission need. Half the time it doesn't, and catching that mismatch in week two instead of week ten is the whole game.
The product owner who used to write user stories from a template now spends afternoons in the field with end users, watching how the current system actually gets used (and worked around). The stories AI drafts from those notes are better than the ones the PO used to write from memory — because they're grounded in observation, not imagination.
In each case, the artifact still gets produced. The person is still essential. But the human's attention has moved up the stack — out of execution, past coordination, into judgment and domain. That's the elevation. It's not a metaphor. It's a different layer of the work.
Who this isn't for
I want to be careful here, because there's a flavor of writing on this topic that pretends every change is a growth opportunity, and that's not honest.
Some people genuinely loved the craft of the routine work. Running a clean standup, building a beautiful Jira board, producing a well-formatted status report — these are real skills, and the satisfaction of doing them well is real. Level up to strategy is not a costless ask. It demands different muscles. Customer-facing time is harder than ceremony-facing time. Domain depth takes years. Sitting in ambiguity is exhausting in a different way than executing a checklist.
It's worth saying out loud: this transition isn't going to feel good for everyone, and that's not a personal failing. Some of the discomfort my mentee was naming is appropriate grief for a kind of work that's genuinely getting smaller.
The window
Here's where I land.
The choice my mentee is facing isn't whether AI takes the routine work. That's already happening, on a weekly and sometimes daily basis. The choice is whether he gets to define what fills the space — or whether someone else defines it for him.
That's a window. It's open right now. The PMs and SMs and POs who are deciding, this year, what their job is for — what their unique contribution looks like when the artifacts are commodity — are going to set the shape of the role for the next decade. The ones who wait will inherit a job description written by people who didn't do the work.
The shift isn't from person doing the work to person not needed. It's from person doing the work to person getting the work done with AI — operating at a higher layer of the stack than you used to, and deciding deliberately what that higher layer is for you.
That's still knowledge work. It might be the most important kind.
Originally published on LinkedIn.