I've been a software engineer for 15 years. I don't write code anymore.
That sentence would have terrified me 18 months ago. Now it feels like the most obvious thing in the world.
The Part Where AI Made Me Worse
I tried GitHub Copilot early on. Fancy auto-complete. And it genuinely made me worse at my job. Not because the suggestions were bad — because of what it did to my brain.
I was so focused on the minute details of what the AI wrote — checking every line, validating every suggestion — that I stopped thinking about the system. How does piece X connect to piece Y in file Z? What API endpoints does this feature need? What should the interface feel like for the user?
I was thinking smaller. My brain was stuck in the weeds, reviewing AI output line by line instead of designing software. I felt dumber. So I uninstalled everything and went back to writing code myself.
The Part Where Everything Changed
In November 2025, Google released Anti-Gravity. Not auto-complete — an actual agent inside an IDE that could see your full project across files, create entire files, edit existing ones, and test its own work.
I opened an empty folder and told it to build me a blackjack game. Detailed rules, betting ranges, multiple hands, the works. Five minutes later, a browser window popped up and I was playing blackjack. The agent had built it, tested it, and launched it. I was blown away.
A few hours later I had baccarat, poker, a slot machine — an entire casino suite. The AI was even generating game art.
Were there bugs? Of course. I'd hit deal before the animation finished and we'd be playing with one card each. But here's the thing — I could look at the code, see the problem, and tell the AI: "chain these animations with callbacks synchronously so one finishes before the next starts." I didn't need to write the fix. I just needed to know what good software looks like.
And I was terrified. Because this felt like the end of my career.
The Part Where It Wasn't
Over a couple months, I started giving agents small jobs on real work projects. I'd describe the data schema, the interface, how things should work — much more detailed than "build me a blackjack game." And the results were good.
But the real shift wasn't in the output. It was in my head.
Instead of thinking about which files I needed to touch, which repositories to work in, which lines to change — I was thinking about the user experience. About what functionality needed to exist. About the system as a whole.
This was the opposite of what Copilot had done to me. Copilot made me think smaller. Agents made me think bigger. With Copilot, I was a code reviewer. With agents, I was an architect.
Then one morning in February 2026, it clicked: "What if I just go all in? I stop coding. I architect systems. I hate coding anyway — but I love creating and architecting systems."
The Part Where I Proved It
I had a 6-year-old ERP/CRM at work. Three business entities sharing one messy codebase, if/else branches everywhere, five different developers' patterns layered on top of each other. I estimated a proper rebuild at one year with two developers. We didn't have the time or the budget.
I told my boss: "Let me spend $200 on AI tools this month. If it works, you get a brand new application for all three entities." He said yes.
One month later, all three entities were live on a completely rewritten, multi-tenant SaaS application. Every feature working. Superior to the previous version.
One developer. One month. $200.
The Part Where People Don't Believe Me
Nobody believes that number. Engineers especially. They think I'm exaggerating, or that the code must be garbage, or that I got lucky.
The code isn't garbage. It's clean, it's standardized, it follows patterns I defined. The AI didn't just vomit out code — I architected every piece of it. I defined the data schemas. I designed the interfaces. I described how every feature should work. The AI wrote the implementation. I tested it. When it was wrong, I told it why and it fixed it.
And then I built an entire agent orchestration system — five specialized AI agents with focused jobs, coordinated by an orchestrator — that now maintains and extends the codebase across five repositories. The whole thing runs on five markdown files.
What I Actually Do Now
My job title still says software engineer. But here's what my actual day looks like:
I think about what needs to exist. I think about how it should work. I think about the data model, the user flow, the edge cases. I document it clearly enough that an AI agent can build it. Then I review what gets built, test it, and direct corrections.
I'm doing the part of software engineering I always loved — the thinking, the designing, the architecting — and I've automated the part I always tolerated — the typing.
The Part You're Thinking
"But what about when the AI writes bad code?"
It does. Regularly. And I catch it because I've been doing this for 15 years. I know what bad code looks like, what bad architecture smells like, what's going to break at scale. That experience didn't become worthless — it became the entire point.
The AI writes code. I know if it's good.
"But aren't you worried about job security?"
I was. In November 2025, I was genuinely terrified. But here's what I've realized: the people who are going to lose their jobs to AI are the ones who only know how to type code. The people who know how to think about systems, who understand why software should work a certain way and not just how — they're more valuable than ever. They just don't need to type as much.
"But you're not really a software engineer anymore."
Maybe not. I don't have a better title for it yet. But I'm building more software, faster, and at higher quality than at any other point in my 15-year career. Call it whatever you want.
The Actual Point
I'm not telling you to stop writing code tomorrow. I'm telling you that the job is changing, and the change isn't what you think. It's not "AI replaces developers." It's "AI replaces typing." The thinking, the designing, the knowing-what-good-looks-like — that's the job now. That was always the real job. We just used to have to do the typing too.
I stopped writing code. I didn't stop building software. I've never built more of it.
If you want to see what that looks like in practice, here's what I've been building: a live band karaoke production platform, a personal AI health system, and the Dark Factory agent orchestration system that's now how everything gets built.