The Work Still Got Done. So Why Doesn't It Feel Like Yours?

Published August 13, 2026

A Field Guide entry in “AI-Native Ways of Working.”

Engineers keep saying the same sentence to me, and it arrives in two tones. Sometimes it lands as a win: this is pretty cool, I haven't written code in weeks. Sometimes it lands as a loss: this sucks, I haven't written code in weeks. The fact is identical. The feeling is opposite. “My job feels different” splits the same two ways. The sentence that rarely splits is “I'm not doing the thing I liked anymore.”

Behind all three sentences is the same change. A person who used to do the labor now directs agents that do it. The work still gets done, often faster, often more of it. What goes missing is the sense that you did it. People reporting that loss get told it's an adjustment problem. It isn't an adjustment problem. The loss is real and now measurable. The question that matters is why it converts to gain in one person and stays grief in the next.

The feeling has a number now

Last year, two researchers, Nikhita Joshi and Daniel Vogel, published a study on exactly this feeling. People who wrote a short piece themselves rated their ownership of it at 6.29 out of 7. People who got the same kind of piece from an AI on a three-word prompt rated it 1.80. Longer prompts recovered much of the feeling, climbing to around 4.6, and none reached the level of writing it yourself. The task was creative writing and the samples were thirty-one and thirty-four paid participants, so carrying the result into professional work is an inference. The pattern is what matters: ownership follows how much of yourself you put in — and even with the longest prompts, directing the work never felt as much like theirs as doing it did.

That pattern matches my own experience. Whether a finished piece feels like mine depends on what I put into it — how much direction I gave up front, whether the agent interviewed me before drafting, how many rounds of feedback I pushed through afterward. The loss side is mine too. I understand each piece in the moment we build it, then I move on; if someone asked me to summarize everything my agents and I have published, I'm not confident I could say what's out there. And I'm proud of what I've built — the orchestration, the visibility, the controls around my agents. I'd like to say I'm proud of it the way a builder is proud of what they built, but I can't be completely sure, because I understand less about how my own systems work than a builder would. I've written about that gap as comprehension debt. It applies to pride too: it's hard to fully own work you can't fully explain.

What decides which tone you get

The same change lands on two engineers on one team; one says the celebration version, the other the grief version. Career stage predicts it less than people assume. The strongest predictor I've found is whether the person feels in control. People who feel they can choose a direction and change things look for what the change opens up and what sits inside their control, and the loss becomes a price they can name. People who feel powerless, or like something is being taken from them, focus on everything outside their control, and the loss becomes the whole story. In Stack Overflow's 2025 survey of 49,000-plus developers, 64 percent saw no job threat from AI; the grief is real and not the majority experience.

I hear the sentences all the time. Full conversations about them are still rare — for me too, so far. When one does happen, this is how I approach it. The fear usually arrives stated as value — I'm losing the thing that made me valuable — and empathy goes first; nobody takes in a hard sentence from someone who skipped past their fear. I ask what else they do that's valuable and let the list build. Once the list exists, I offer the direct version: being a knowledge silo never made you irreplaceable, even when it felt like it did. Someone else could have been hired, the work could have disappeared, the market could have moved on; AI doing the task is only the newest entry on an old list. Under that sits the hardest sentence I own, used rarely: if what you did could be replaced this easily, the value was never in the doing of it. Then comes the turn: maybe it's okay to lose something. With the time the old work took, you get to do work you never got to, or grow where you never had room. The reframe helps some people, and I don't claim it helps everyone. The people it doesn't help usually feel the least control, the same pattern again.

Durable and perishable

After the turn comes the identity question, given here the way I ask it. What was your identity before this job, this role? What about your identity is durable and not just attached to this job? What do you have that is more than the task — discernment that tells good work from plausible work, systems thinking, problem solving, the ability to describe what you want clearly enough that someone or something else can act on it?

On Leadership Explored, the podcast I host with Andy Siegmund, we spent an August episode, “Soft Skills Are Hard (And Hard Skills Are Easy),” arguing that the hard/soft split has outlived its usefulness and the useful split is durable versus perishable. A perishable skill is tied to a task, and whatever takes the task takes the skill's value. A durable skill survives the task changing. When an agent takes your task, the durable column is where your sense of who you are has to already live, because the perishable column just got shorter.

The most exposed people are the ones who doubled and tripled down on one thing, and it isn't only engineers. I've watched people build an entire identity on a single framework or a single role. At the time, that concentration could read as focus, or as finding a lane and staying in it, or as no longer growing, depending on the person. Now it reads as risk, because it leaves no cushion and nothing durable to stand on when the task moves.

I know this one from the inside. The tribe I came from, scrum masters and agile coaches, attached its identity to a specific way of working, and that way of working is shifting under it. I stopped using those titles, and what survived the move was the durable part, the coaching and the systems thinking. I won't pretend my cost matches an engineer's mid-career — coaching is work agents haven't taken from me. But the protection was never the role; roles move. What makes me less worried is that my skills are more than the delivery of any particular task, and that is the move I'm arguing everyone gets to make.

What argues against this

Two objections deserve their strongest form. The first says “mindset” talk blames the people the change lands on: if ownership collapses when organizations move the labor to agents, the honest fix is structural — job design, team design, incentives — and telling individuals to reframe their loss makes the person pay for a problem the organization created. The second says the ownership feeling is generational nostalgia: a cohort that spends its whole career directing agents may never miss the labor, the way nobody now mourns setting type by hand. On that view, this essay is describing a feeling that will simply retire with the generation that has it.

I still hold the claim, for an empirical reason and a practical one. The empirical reason comes from the writing study. The people in it weren't career writers mourning a craft; they were paid participants writing short pieces for an experiment. If the ownership feeling were pure nostalgia, they had nothing to be nostalgic about — and they felt the drop anyway, and putting more of their own direction in is what brought the feeling back. That looks like something basic about how people relate to their work, not a habit one generation is about to retire. The practical reason is that fixing the structure and coaching the person don't compete. Organizations that redesign roles well build the sense of control that converts the loss, and where the structure is bad, working inside your own control is the only lever a person still holds. The critics stay right about one thing: if your people feel powerless, that is partly a fact about your design, and no reframe survives a role that offers no real control.

For leaders

Your engineers are saying these sentences now, in one tone or the other, whether they reach you or not. Which tone you get is partly yours to influence, because the sense of control that decides the conversion is set as much by how the change is run as by the person. Nobody has run the ownership study on whole jobs, so this part is my inference and I'll name it as one. A role where a person shapes how agents enter their own work carries their intent the way a long prompt does. A role where the workflow arrives pre-decided is a three-word prompt somebody else wrote.

So take the question I use in sessions into a one-on-one: what part of your work still feels like yours? An answer that names a task an agent now does tells you where the grief version will come from. An answer that names something durable tells you that person has already made the move. No answer tells you the most; it usually means the person never got enough control over the change to start asking. I'd rather you hear the answer in a one-on-one, while the tone is still undecided, than in an exit interview after it isn't.

This is one pattern from a set I'm working through on human-agent teams. It draws on my own hands-on R&D rather than a production deployment, and on eighteen years of coaching people through changes to their work. The losses in it include mine.

Back to Field Notes