Something something – A Paradox

I guess an LLM would title this “Something something – A Paradox”

True: Current AI is incredibly useful and powerful.

True, but probably more interesting: It is also utterly and COMPLETELY UNUSABLE for the regular knowledge worker (read: you are not a software engineer).

I work with it every day. I am a curious autodidact with a decent technical understanding and high tolerance for FAFO. Plus, I have people that I can reach out to and ask questions and get great answers (my questions are highly embarrassing but my curiosity drive wins here).

If you have that exact combo, congratulations!

You can now push agents around, or getting pushed around and prompted by them.

You have more than five, and usually not more than 25, things going on somewhat concurrently (that is what tech people say for at the same time).

You cycle through your sessions, lanes, work streams and projects, saying OK, or picking multiple-choice type pre-written answers, or correcting, or to copy and paste a command into another terminal window, or reading a report, or making decisions, or trying to decode the cryptic concoctions of Opus 5 (until you ban it from your list of delegates).

You manage your agents’ context while having to switch context yourself at a rapid pace.

It is not one app, or two, but more like four to five surfaces, as we’re now calling them. You have to absorb switching those, too. It’s just what it is.

That’s not all.

Since this is AI and it can write code and all, you are also continuously developing and improving your setup and your tooling.

In the time it would take me to talk you through my setup, it has already changed and is now outdated. Just how it is. There are new models being released, new features in each of the apps, new skills, tweaks to AGENTS.MD and other shenanigans. I mean, reading all the relevant changelogs alone would take half a day each week. And then what about the research papers, the hardware releases, the growing number of security breaches and all the commentary on it?

Before I 10x’d my productivity, back in the olden days, I had time for that. Now I read verbose AI messages until my eyes bleed.

P.S.: Yes, no, I did not draw this graphic by hand.