AI Is Coming for the Part of Your Job You Were Proud Of

The comforting take you've read a hundred times goes like this. AI isn't coming for your job, it's coming for your tasks. The boring repetitive stuff gets automated. You get to focus on higher-value work. Everyone wins. Net gain of 78 million jobs by 2030. Relax.

That framing is technically correct and deeply misleading. It's missing the part that actually hurts.

AI isn't coming for the boring tasks. It's coming for the exact tasks you spent a decade getting good at. The ones that you used to prove you were a senior. The ones you'd point to when a junior asked "how do you know so much?" The craft. That's what's getting flattened, and the framing of "just the routine stuff" lets us avoid admitting it.

What Happened to My Craft

Here's the concrete version of this for me. I have been shipping code for twelve years. For most of those years, my edge was a stack of skills that took me a decade to accumulate: fluency in four or five languages, an instinctive sense of which ORM queries would explode at scale, a working memory of thirty or forty framework APIs, the reflexes for debugging weird network issues by pattern-match, and a library of small snippets of code I could recite from muscle memory. If you asked me what I was good at, those are the things I would have named.

All of it is now a commodity.

Claude Code wrote most of the code in this project. When I hit a weird bug yesterday in our X API integration, I didn't reach for my debugging reflexes. I described the symptom and dispatched a research agent. When I needed to add a new feature to our Next.js app last week, I didn't open the Next.js docs, I just told Claude what I wanted and reviewed the PR. The muscle memory is still there, I can still do all of it by hand, but it doesn't matter anymore. The hand-written boilerplate that used to mark me as senior is now a checkbox.

The boring tasks were not what got automated. What got automated was the craft. The thing I was proudest of. The thing that felt like proof I knew what I was doing.

This is a strange feeling to process and most posts about "the future of work" skip past it because it doesn't fit the optimistic narrative. But you cannot honestly talk about AI and jobs without admitting that the losses here are not evenly distributed. The person losing the most isn't the one doing data entry. It's the craftsperson who spent years becoming excellent at a thing that just got easy.

Most People Think the Comforting Take Is Kind

Most people think the "tasks not jobs" framing is kind, because it sounds less scary than "AI is going to eliminate your job." But it's a worse kind of cruelty: it lets the person on the receiving end avoid processing the actual loss. Their job is "safe." What's actually happening is that the skill stack they built their professional identity around is rapidly becoming a commodity, and nobody is telling them that out loud because telling them would be rude.

The correct take is harder and more useful: AI is coming for the part of your job that made you special. Now you get to decide what kind of person you want to be on the other side of that.

What's Left When the Craft Goes?

This is the real question, and it's not "learn Python and prompt engineering." The real question is: when execution gets cheap, what does value look like?

The answer seems to be judgment under ambiguity. The specific kind of judgment that comes from years of being in rooms where things went wrong, and knowing what "wrong" felt like before the metrics showed it. Knowing which feature is worth building and which is a trap. Knowing which client is about to churn. Knowing which code change is "fine to ship at 5pm on Friday" and which is "absolutely wait until Monday morning with backups." Knowing when to abandon a plan.

None of those are tasks. You cannot put them in a Jira ticket. They don't show up on a resume. They are the thing experience actually buys you, and they are also the thing most people don't consciously develop because, up until now, the craft stack was enough to get you paid.

The people who will thrive in the next two to three years aren't the ones who "use AI daily." They're the ones who quietly realize that their craft is commoditized and pivot hard into being the person in the room with judgment. This is harder than it sounds because you cannot list it on LinkedIn. But I think it's what's actually going to separate the winners from the losers.

The Prediction

By end of 2026, "seniority" in software engineering (and arguably in most knowledge work) will stop meaning "years of experience with a technology" and start meaning "quality of judgment under ambiguity." Job descriptions will start to shift. Interview processes will start to shift. The question "have you used X framework" will matter less and less. The question "describe a time you made a non-obvious call under uncertainty and how it played out" will matter more and more.

I'm confident enough about this to say: if you are in a hiring role and you are still screening candidates primarily on their fluency with specific tools, you are going to make increasingly bad hiring decisions. The most valuable senior people right now are the ones who make tools irrelevant because they adapt instantly to whatever tool is in front of them. That's not a skill. That's a way of being, and you hire for it very differently.

Screenshot this post. By Q4 2026, if senior engineering interviews at well-run companies haven't shifted noticeably toward judgment questions and away from framework questions, I was wrong. I don't think I'm wrong.

Who's Actually Better Off?

The second piece of the optimistic take I want to push back on: the idea that junior workers are somehow in worse shape than senior ones. I think the opposite is true right now.

A junior today doesn't have a craft stack to lose. Their professional identity isn't built around remembering framework APIs or writing boilerplate. They can skip directly to the judgment-building phase by using AI to handle everything below judgment. I watched a 22-year-old last month ship a full-stack app in a week that I could not have shipped as a 30-year-old working hands-on, not because she was more skilled than I was at that age, but because she was not carrying the cognitive load of "do I remember how to write this by hand." She was spending 100% of her attention on the question "is this the right thing to build?" which is the actual high-value question.

Meanwhile, the senior engineer who came up writing hand-optimized SQL and now has to watch Cursor do it faster is having an identity crisis they didn't ask for. The craft they built their confidence on is evaporating, and the thing that's supposed to replace it, judgment, is a thing they have, but it's not what they were trained to lead with.

If you're junior, lean in fast. If you're senior, the work is harder: you have to let go of the craft you earned and deliberately rebuild your professional identity around something less visible. It's the harder pivot, but it's the one that's available, and everyone I know who has made it comes out stronger on the other side.

The Honest Version

AI isn't coming for your job. It's coming for the part of your job that made you feel skilled. Your job will still exist. It'll just pay you to do a different kind of work than the one that made you proud.

The people who do well will be the ones who decide, on purpose, to grieve the loss of the craft and get on with the rebuild. The people who don't do well will be the ones who insist "AI is coming for tasks, not jobs" and then quietly wonder why their last year at work felt so hollow.

The boring tasks were never the prize. The craft was. The craft is what's changing. Be honest about it and start planning accordingly.