Two tweets crossed my feed this week, and both hit a nerve.
Klaas wrote that software engineering is now draining because all the "easy" work is done. You used to have days of productive but relatively simple work — writing some new endpoints, making a dashboard. Now all that's left is the hard part.
Dillon Mulroy wrote that it's way harder to get joy and satisfaction out of building with AI than it was before. Constantly straddling burnout, far less immersed in hard problems, constant context switching. Tired of the uncertainty of where this is going and how to do it well.
Hundreds of thousands of views on each. A lot of people feel this.
I recognize it too. And I want to be honest about it, because my answer is not "just prompt better".
I'm more productive than I've ever been
Let me start with the good part.
Last week I shipped a new product from zero to production in 6 days. I built over 200 free tools for this site. I'm running multiple projects at once, alone, and moving faster on each of them than I used to move on one.
None of this would exist without AI agents. Some of it I would have never even attempted by hand.
So no, I'm not going back. That option is gone for me.
But something did change
Klaas is right about the easy work.
Writing a CRUD endpoint, wiring up a form, building a settings page. That work was never intellectually hard, but it was satisfying. You sat down, you typed, an hour later something worked. Small win, dopamine, repeat.
That loop is what a lot of us called "the joy of programming". And agents eat exactly that work. They're great at it. It's the first thing they got good at.
What's left for you is the part that was always draining: deciding what to build, judging tradeoffs, reviewing, catching subtle problems, saying no. All day. Every day.
Dillon's version is the workflow side of the same problem. If you run five agents in parallel, you're not programming anymore, you're air traffic control. No immersion, no flow, just context switching until your brain gives up.
Do I miss writing code by hand?
This is the part I keep going back and forth on.
I don't know if writing code by hand is even worth doing anymore. I genuinely don't. Some days I think it's like insisting on developing your own film after digital cameras arrived. Other days I open an editor, write something small myself, and remember why I started doing this in the first place.
I don't have a clean answer, and I'm suspicious of anyone who claims they do. This whole thing is a few years old. We're all figuring it out in public.
Where the joy went, for me
What I can tell you is what happened to my own joy. It didn't disappear. It moved.
It moved from writing code to deciding what exists. The satisfaction I used to get from a function coming together, I now get from an idea becoming a real thing people can use, the same week I had the idea.
It moved from implementation to shipping. I ship constantly now. Shipping was always the best part, and I get a lot more of it.
And the craft didn't die, it changed subject. I used to care about clean code. Now I care about clean products. Taste still matters, maybe more than before, it's just applied one level up.
Two things protect me from the burnout Dillon describes:
I mostly single-task. One agent, one problem, my attention on it. Running many agents in parallel looks impressive and feels terrible. I do it sometimes, for mechanical work, but it's not my default. The productivity gain is not worth feeling like a dispatcher.
I build useless things for fun. This is what my Summer of Code is: no launches, no roadmap, just building what I want. Some of it with agents, some of it by hand, because I feel like it. Nobody said your hobby has to be optimized.
The open question
Maybe the honest framing is this: AI didn't remove the joy from programming. It removed the joy from a specific kind of programming, the kind that happened to be most people's day job.
Whether the new kind — directing, reviewing, deciding, shipping — can be joyful too, or whether it's just management with extra steps, I think depends on the person. For me it mostly works. Clearly for a lot of people right now, it doesn't.
If you're in the second group, I don't think the answer is to quit, and I don't think it's to pretend nothing changed. Maybe it's to protect a small space where you still write code because you want to, not because it's efficient.
I'd genuinely like to know where you landed on this. My inbox is open.