I’ve been using LLMs to write all my code since July 2025, but the first few weeks of March saw something strange happen. I woke up at about 4am, and I worked until at least 4pm - often 6 or 7pm - taking breaks only for meals (and perhaps a call or two). That’s a VERY large amount of work! And my brain is so tired at the end of it. But… I wanted this. #
Yes, I think the nature of software engineering has changed. If you’re managing LLMs who write code for you instead of doing it by hand, you’re playing a very different game with your brain chemistry. You’re no longer spending three hours getting lost in a problem statement and getting into flow; you’re managing 6 different contexts, switching between them in the downtime of the others. You become a manager instead of a creator. You become a slave to making sure the machine has enough to eat. #
After using Claude for 6 to 10 hours a day, my brain feels the same way it did in 2018 and 2019, when I was doing Not-Engineering for work, having to manage vendors and customers and research and strategy and following up with about 80 different people and staying on top of the trends so I can capitalise on them in some way. Each conversation checkpoint feels like reaching the end of a long comment thread on an entertaining AskReddit. Each time I come back to the same tab containing a slightly different context the brain signals travel through the same pathways that coming across a *different* short from *Gray’s Anatomy* does. #
After a long, hard, day at work, I come home exhausted. I cannot consider socialising with anyone because I am out of words. I take the time to wind down. I feed the cats, do my skin care, water the plants, prep food for tomorrow. I take the cats through playtime, shower, and head to bed… where my brain, after having been on the hamster wheel for ten hours and then having three hours of rest, is ready to go again. It’s horrible. My cortisol is elevated, my attention split, and I feel like I’m always dropping a ball I don’t yet know about. #
Three years ago, when I realised how much social media usage was eroding my psychological sovereignty, I was able to simply quit. But I cannot simply quit my job - more importantly, I don’t want to. I enjoy writing code, but only in as much as it is useful to me to get things done. As someone with high agency, LLM-backed code generators have thrown open problem spaces - from things I theoretically know is possible but don’t have the time to do anything about, to things I can actually build, given two days and several million tokens. #
The cost of this is that I need to figure out how to protect my brain the way people with social media jobs need to learn to protect theirs. #