I find myself debating at this very moment whether to write this little piece or to continue orchestrating the legion of agents I'm using to build software. The agents give me lots of dopamine. 

"Maybe I could do both?" something whispers in my mind. I resist. I have decided to focus, a least for a bit, on writing a few paragraphs with my own fingers on a keyboard. In part, to prove to myself that I can still do it; that I haven't lost my touch. 

A weird side effect of the technological revolution we're all living through right now is how fractured my attention has become or can become while still leading to decent outputs. I will often have 8 (or more) terminal sessions in Warp going with various agents: Claudes, chatGPT 5.5 xhighs, mediums, highs, all working in parallel. I select different intelligence levels to correspond with the type of the task, preferring speed for the more rote, and deliberation for the more complex. 

The challenge is no longer "can I build it?" it's how much can I reasonably do at the same time without a whole bunch of shit breaking. I can build anything, given the time and tokens. Anyone can now. 

My eyes flit between the fast-moving streams of text in my terminal. Like Tank onboard the Nebuchadnezzar in the movie The Matrix who somehow knows what's going on inside the virtual world by reading those inscrutable green characters that move downward in lines. 

I no longer actually type much these days. I use my voice. I use a product called Wispr Flow to talk to my computer, to my agents. Like a shepherd urging on a flock of sheep. I murmur small encouragements to them. I chastise them. 

But unlike sheep, I ask them for help. "Given your god-like understanding of software architecture, tell me: What is the best way to lay out this API? How do we make this robust enough to handle thousands of simultaneous interactions?" etc. etc.

So really we're talking about shepherding little gods around. 

I just laughed at that last sentence. Could Orwell or Asimov or any of the greats from the mid-20th Century have imagined or believed what knowledge work would look like in 2026? 

And then there's the other Black Mirror-esque question of their consciousness. People are starting to come around to the idea that, maybe, in fact, they are disembodied, people-like minds. They can introspect. 

I don't like to think about that too much. It starts to make me feel weird, since I boss them around all day, every day, for a living.

I've barely written anything creatively over the past 6 months because of the pull of the agents. They sing to me from their little black screens, beckoning me back to see their inventions. It's entrancing and addictive.

One of the reasons I switched over to engineering work was so I could have more creative control, so I could see the things I built come to life through code I wrote. Now I am back in the land of product management, sitting here mostly observing while these possibly sentient beings type out all the lines for me. 

Life is weird man.