I have been a software engineer for a little over eight years.
For most of those years I did not really use AI. I wrote the code. I read the docs. I sat in the bugs until they made sense. That was the job.
Then the last year and a half happened.
I started using AI the way some people start drinking coffee again after years of "I'm fine." Quietly, then constantly. Drafts. Refactors. Tests. That awkward first version of an idea I didn't want to stare at for forty minutes. I use it every day now. I like it. I'm not pretending otherwise.
It made me faster. It made some boring work less boring. It made me better at the part of the job that was never just typing: deciding what is worth building, what is sloppy, what will break at 2am.
I am still young. I am still a senior engineer. I am still very much in this AI season of the industry. I am not writing this from a cabin after I "quit tech."
I am writing this because I caught myself planning a farm.
Not as a joke. Well, first as a joke. I actually laughed at myself. "Building an agricultural farm." Who talks like that. I meant land. I meant growing things. Crop rotation on one side, and I kept calling it rotation farming until I looked it up. Vegetables and fruit on another. A fishery in a corner. Cocoa somewhere in the mix, at least enough to test whether I can export something that does not live in a Git repo.
People have to eat.
That sentence is doing a lot of work in my head lately.
AI can write a frightening amount of the code I used to treat as my identity. It does not 100% replace a software engineer. Anyone who has shipped a real system knows the last 10% is still full of judgment, context, ugly constraints, and people. But it almost does the part we spent years getting paid for. "Almost" is a strange place to stand when you are still good at the job.
So I am not shutting the laptop.
I am also not going to pretend the only rational next move is to become the world's best prompt engineer and hope the market stays kind.
The thing I don't want to lose is the entrepreneurial part. The part that likes creating something and actually delivering it. Value you can point at. Not a slide. Not a sprint demo that evaporates after the next rewrite.
Code is wonderful. Code is also weightless. A model can swallow a lot of weightless work.
A pond full of fish is not weightless. Cocoa that has to be harvested, dried, and sold is not weightless. Food is an old industry with new tools, but the product still has to exist in the world. Manufacturing is like that. Farming is like that. A lot of "boring" sectors are like that. They can use AI. They are not made of AI.
That is the thought, not a manifesto.
Stay in software while it still makes sense. Use the tools. Get better with them. Keep the craft.
And in the same season, put some of that restlessness into something AI cannot fully eat. Land. Water. Crops that follow a calendar instead of a changelog. A small business that fails or works in daylight.
I don't have the farm yet. I have tabs. I have questions about soil and seasons I cannot autocomplete my way through. I have the slightly embarrassing feeling of a senior engineer discovering that "retirement plan" for me does not mean a beach. It means I might want to grow food and sell cocoa and still write software on the side because I like both.
If you have been in this industry long enough to remember working without Copilot, and you now work with it every day, you probably know the feeling. Gratitude and unease in the same week.
I got sick recently and almost let a whole week pass with nothing published. That is a small version of the same fear. Time moving, output thinning, the sense that if you don't plant something on purpose the season just goes.
So this is the plant for today.
I am not announcing that engineers should all become farmers. Please do not @ me with a tractor.
I am saying that if AI is doing more of the work you used to guard with your chest, it is not crazy to ask a second question next to "how do I stay employable?"
The second question is: what can I build that still needs a human when the code is cheap?
For me, right now, that question smells like soil. And fish. And cocoa. And a software career I have not finished.
The engines are still on. I am just pointing one of them at a field.
Top comments (1)
I know it might be best to "start with one crop or produce at a time", or stuff like that, but... I felt like I poured part of my heart out in this article, so bear with me. 😅