The Hidden Duty That’s Burning Out the Middle
Somebody has to watch the machine.
That job did not show up on the position description. It does not have a line on the performance plan. It just landed on the people in the middle, the ones who used to turn messy work into something a decision-maker could use.
Now they spend a chunk of every week doing something the surveys call botsitting. Checking what the tool produced. Filling in the context it never had. Catching the confident wrong answer before it leaves the shop. Rerunning the prompt. Cleaning the draft. Explaining to the person above them why the “finished” product is not finished.
Open-source workplace studies put that hidden labor in the range of a full workday each week for people who live in these tools. Almost half the time spent “using AI” is not creation. It is supervision. The tool looks fast. The human in the middle absorbs the delay.
I’ve watched it in real rooms. The mid-level person who used to come in with a point of view now comes in with a packet they barely trust. They talk a little faster than usual. Eyes bounce between the slide and the boss. If you ask a question that is not on the page, there is a pause. Not a thinking pause. A search pause. My Profitable Body Language book treated that kind of incongruence as a hotspot. The words say the analysis is ready. The rest of the person is still babysitting the source.
Sun Tzu was ruthless about this kind of drain. He did not measure an army by how many banners it could raise. He measured whether the force still had strength when the real fight started. An army that looks busy marching in circles is already losing. He also understood that the officers in the middle hold the line between the plan and the field. Cut them, overload them, or turn them into full-time inspectors of junk, and the general at the top starts making decisions on fog.
This is different from workslop, which is the junk itself. Botsitting is the duty of catching it. One is the product. The other is the tax. And the tax is landing on the layer organizations are also trying to thin out. Fewer managers. Wider spans of control. Same delivery dates. Add a stack of tools that need watching and you did not get efficiency. You got a smaller group of people doing two jobs, one of which nobody will admit exists.
Manager engagement has been falling in the open data for a couple of years. That tracks with what you see if you sit in enough mid-level meetings. The people who used to coach, prioritize, and protect the team are now quality control for software that does not get tired and does not know when it is wrong. Some of them start signing off just to keep the queue moving. That is how you get the next problem. Once the middle stops checking, the slop goes straight to the decision.
I hit a related idea in my Redirecting Rejection book. The moment you treat every incoming demand as something you have to white-knuckle through, you lose the chance to redirect it. Botsitting works the same way. If the only available move is “fix it and send it up,” the load never changes. The move that actually works is naming the duty out loud and deciding what is worth watching.
If you lead these people, stop pretending the tool removed work. It moved the work. Ask what they are actually babysitting this week. Then cut something. A status meeting. A second tool. A report that exists because the first report was slop. Sun Tzu did not keep every outpost manned because it looked impressive on a map. He kept the ones that mattered and let the rest go.
Give the middle a standard for “done” that a human can defend. If they cannot explain the second paragraph without going back to the prompt, it is not ready. That sounds obvious. It is not how most shops are running right now. Volume is still winning the argument.
If you are the one botsitting, treat it as terrain, not a character test. You cannot outwork an infinite generator. Protect the hour you need to think. The general who never leaves the inspection table never sees the rest of the field.
The organizations that stay sharp will not be the ones with the most agents running. They will be the ones who still have a middle that can tell the difference between watched work and owned work. That layer is not overhead. It is the only thing standing between a clean-looking briefing and a bad decision.
Stay Strong, Stay Disciplined, Stay Dangerous,
–
S