Most of us calculate AI's value the same way. We look at the task it used to take an hour and now takes ten minutes, and we call the other fifty minutes saved. That math feels obvious. It is also incomplete.
What we are actually seeing, once teams track a full week instead of a single task, is a second column showing up next to the time saved. Recent workplace data puts it at something like six hours a week spent checking AI output, fixing what it got wrong, and rerunning prompts that did not land the first time. People are starting to call this botsitting, and once you have a name for it, you start noticing how much of the week it quietly eats.
That matters for time because the win we are counting at the point of creation is not the win we are keeping. A chunk of it gets handed straight back to review, correction, and re-generation, and if we are only measuring the first ten minutes, we are missing where the rest of the hour actually went.
------------- Context -------------
Right now, most teams still budget AI the way we budget a faster tool. We assume the task shrinks and the freed-up time is ours to redirect toward something better. That assumption drives a lot of the adoption enthusiasm we see across the community: faster drafts, faster summaries, faster first passes.
But a faster first pass is not the same as a finished one. AI output that looks complete can still be wrong in ways that are easy to miss on a skim and expensive to miss on delivery. Wrong tone, wrong numbers, a missed nuance in a client email, a generated summary that quietly drops the one caveat that mattered. None of that shows up as an error message. It shows up later, usually after someone has already trusted it.
This is where botsitting becomes a useful frame rather than just a complaint. It names the specific work of supervising a fast but unreliable collaborator: check the output, catch what is off, decide whether to fix it or start over. That is a real task with a real time cost, and it deserves its own line in how we plan our week, not a footnote.
Once we see it that way, the deeper issue becomes visible. We have been budgeting for creation time and treating supervision time as incidental. It is not incidental. For many recurring tasks, it is now the larger half of the job.
That is a time issue because until supervision gets budgeted on purpose, it keeps arriving as a surprise, and surprise time is the most expensive kind. It shows up as a missed deadline, a rushed review, or a task that quietly takes as long as it always did, just with an extra step bolted on.
------------- The Time Saved At Creation Is Not the Time You Keep -------------
One of the quietest costs in AI-assisted work right now is the gap between the time a task appears to save and the time it actually saves once review is included. A ten-minute draft is genuinely a ten-minute draft. It is not, by itself, a finished piece of work.
This is where the botsitting data becomes clarifying. Recent workplace research found AI users saving roughly eleven hours a week while spending upward of six of those hours checking, correcting, or rerunning the very output that created the savings. That is not a rounding error. That is more than half the win going straight back into supervision.
Consider a consultant who used to spend forty minutes drafting a client update by hand. AI now produces a passable draft in six minutes. But the consultant still spends eighteen minutes checking figures, softening a tone that reads too blunt for this particular client, and rewriting one paragraph that missed the actual point of the update. Total time: twenty four minutes. Real savings: sixteen minutes, not thirty four.
That sixteen minutes is still worth having. But it is a different number than the one most people are mentally banking when they say AI cut their draft time by eighty percent. The gap between the two numbers is where a lot of quiet frustration comes from, the sense that AI should be saving more time than it feels like it is.
The direct time gain here is real, it is just smaller and less even than the headline number suggests, and knowing that in advance changes how we plan around it.
------------- Unbudgeted Supervision Time Becomes Invisible Overtime -------------
When supervision is not planned for, it does not disappear. It gets absorbed, usually into the parts of the day that were supposed to be free.
We are starting to see this pattern repeat across the community: someone automates a task, feels an initial rush of speed, and then finds themselves checking AI output during what used to be a lunch break, a commute, or the last twenty minutes before a deadline. The task got faster. The person's day did not get lighter, because the review work moved into the margins instead of the schedule.
A small business owner running proposals through AI might generate a first draft in five minutes, a task that used to take forty five. That looks like a forty minute win. But if the review, fact check, and personalization pass gets squeezed into whatever gap appears next, that time is still being spent, just without being counted anywhere. It shows up as a slightly longer day, not a shorter one.
This is a direct time gain turning into invisible overtime. Forty minutes of drafting time genuinely disappeared. But without a planned slot for review, an equivalent chunk of time reappears somewhere else in the day, usually somewhere less visible and less protected.
That matters because unplanned time is the first thing that gets sacrificed when the day gets busy, which means the review step that actually catches mistakes is exactly the step most likely to get rushed or skipped under pressure.
------------- The Fix Is Not Less AI, It Is a Named Review Slot -------------
The instinct when supervision time creeps up is to blame the tool. Fewer AI outputs, tighter prompts, more careful setup. Some of that helps. But it treats the symptom rather than the actual gap, which is that most workflows have a creation step and no matching review step.
Teams that are getting real value here are not the ones using AI less. They are the ones who have given botsitting an actual slot in the process, the same way a proposal has always had a proofreading step or a report has always had a sign off. The task is not new. What changed is how much of it now needs to happen and how quickly it needs to happen after generation, while the context is still fresh.
A content team that treats AI drafting and human review as two named steps, each with its own time allocation, ends up with a more honest schedule than a team that treats AI as a single step that happens to include some checking. The first team can tell you how long a piece actually takes. The second team keeps being surprised.
This is a direct time gain because named time is protected time. A ten minute review slot that is actually scheduled gets done properly. A ten minute review slot that is squeezed into whatever gap is left gets rushed, and rushed review is where the expensive mistakes slip through.
------------- Practical Moves -------------
First, track one recurring AI-assisted task for a week and log both the creation time and the review time separately, because most people are surprised by how close the two numbers actually are.
Second, give review its own slot in the workflow rather than treating it as something that happens automatically after generation, the same way a draft has always had a dedicated proofreading pass.
Third, sort your recurring AI tasks by how expensive a missed error would be, and put your tightest review discipline on the ones where a mistake reaching a client or customer costs the most time to undo.
Fourth, resist judging AI's value by the speed of the first draft alone, and instead measure the full cycle from prompt to finished, reviewed output.
Fifth, if botsitting time keeps growing on a specific task, treat that as a signal to fix the prompt, the template, or the source material feeding it, rather than accepting the review burden as fixed.
------------- Reflection -------------
The botsitting data is a signal, not a complaint about AI. It tells us that the honest measure of time saved is not what happens at the moment of generation, it is what happens across the whole cycle from request to trusted, finished output.
That is why naming and budgeting review time matters so much. It turns a hidden cost into a planned one, protects the review step that actually catches mistakes, and gives us a real number for what AI is saving instead of the flattering one we tend to calculate by accident.
In the end, the teams getting the most out of AI right now are not the ones generating the most output. They are the ones who have stopped being surprised by how much time supervision takes, and built a schedule that has room for it.
Where in your own week is botsitting time currently hiding, in a lunch break, a commute, or the last twenty minutes before a deadline?
What is one recurring AI task where you have never actually measured the review time, only the drafting time?
If you gave supervision its own named slot tomorrow, what would you learn about how much time a task really takes?