My brain is rotting because of Claude Code and the pressure to

I used to be one of the skeptics, the person holding out as long as possible before adopting these tools. But the threat wasn't theoretical. I was eventually told, quite clearly, that if I didn't "improve my productivity to stay on par with my colleagues," my position would be at risk. That was the moment my descent into an AI-driven workflow began.

Initially, I treated Claude Code like a junior developer. I would delegate small, isolated tasks, but I remained the gatekeeper. I reviewed every single line of code, opened formal pull requests, asked follow-up questions, and forced it to refactor anything that didn't meet my standards. I was still in control.

Then the comparison trap set in.

I noticed my colleagues were shipping code at a rate that felt impossible. To keep up, I started cutting corners on my reviews. I moved from reviewing small features to pushing them directly to the main branch. To match their 10x output, I leaned harder into the agentic workflow. While the agent spent 20 or 30 minutes processing a task, I found myself mindlessly scrolling social media—a bad habit I had fought to break for years.

But even that wasn't "productive" enough. My peers weren't just using one agent; they were using work trees and running multiple LLM agents in parallel. So, I scaled up. Instead of scrolling, I started launching four or five agents at once, juggling their various outputs.

The cognitive load eventually hit a breaking point. I reached a stage where I simply didn't have the mental capacity to track all that context anymore. When the agent presented a solution, I stopped analyzing it. I just started clicking the "Recommended" suggestion.

Now, the situation is dire:

• Code Ownership: I no longer actually know the codebase or how the underlying logic works.
• Debugging: When a bug report comes in, I don't investigate. I just copy-paste the user's complaint into the prompt and let the agent figure it out.
• Research: The ability to deep-dive into documentation or search for edge cases is gone. I just let the AI do the research.

The terrifying part is that it actually works. The code ships, the features land, and the metrics look good. But I feel like a mindless blob. I’m struggling to even write this post because I’ve lost the habit of thinking through syntax, punctuation, or even spelling. The AI is so good at interpreting my "messy" input that my brain has stopped trying to be precise.

People tell me to "just write code by hand" or "quit and find a company that values craft." But in the current market, that feels like professional suicide. Every interviewer I meet asks how I integrate AI into my daily workflow, or they tell me, "We move fast here, so we use AI as much as possible."

I’m stuck in a loop where I’m working 10-hour days managing 15 parallel agents just to pay a mortgage, all while feeling my ability to think critically vanish. It’s a strange, hollow kind of exhaustion.