Your Managers Need More Than AI Training
Prompt skills are the easy part. The capability that makes then pay off sits one layer above the tool.
“How do we up-skill our teams for AI?”
I am hearing some version of that question in nearly every executive conversation right now. It is the right question, the budget behind is justified, and the leaders asking it are ahead of the ones who are not.
Here is the part that rarely gets examined. That training is going to land in a working day nobody has looked at.
What does that day actually hold?
Meetings from morning until the end of the afternoon, most of them back-to-back. No time to prepare for any of them, so nobody does. The real work starts after the kids are in bed, and the deck goes out at midnight.
The next morning arrives on six hours of sleep. The decisions made that day are not as good, so a week later somebody reopens one of them. Reopening it takes a meeting, and that meeting takes an hour out of a day that did not have a spare hour in it.
Then it happens again the following week.
That is the part I want you to sit with. Every piece of that week creates the next piece. The unprepared meeting produces the weak decision. The weak decision produces the rework. The rework fills the calendar that made the meeting unprepared in the first place. Give somebody a one-day course and send them back into that same week, and the week wins.
Give somebody a one-day course and send them back into that same week, and the week wins.
None of this shows up in a prompt engineering course.
Why has unclear direction never cost us before?
Give a colleague a half-formed instruction and watch what happens. They ask you a question. Or they make a sensible guess and check it with you later. Or they quietly rewrite your vague ask into something they can work on.
That repair is invisible and unpaid, and it has been covering for the thin direction in every organization I have ever worked in.
There is a reason managers do not notice they are relying on it. Once you know what you meant, you cannot hear how little you said. Researchers call this the curse of knowledge, and it is stubborn, because the information you left out still feels present to you.
So, most managers have never had honest feedback on how they hand off work. Somebody else has always closed the gap for them.
We have run this experiment before. Organizations trained thousands of people on agile ceremonies and far fewer on what makes agility possible: deciding without complete information, changing course without treating it as failure, saying the difficult thing early. The ceremonies arrived on schedule. The behavior did not. Which is how we ended up with stand-ups that re status meetings and retrospectives where nobody says what they really think.
Now put three agents in that meeting.
Same meeting as last week. No agenda. Somebody asks what we are deciding today, and when no answer comes, the leader says let’s get the agents to pull some options together.
Agent one returns forty pages in ninety seconds. Agent two got a slightly different version of the same request and returns forty pages arguing the opposite, with sources. Agent three was told to make sure everything lines up, and has reconciled the first two into a single confident recommendation that is wrong in a way nobody will catch for six weeks.
Nobody reads any of it. The next meeting starts in four minutes.
The following week, same meeting, no agenda, three more documents.
Then the notes come back from the AI notetaker, and they are in two lines long. The tool did not fail. There was nothing to record.
The tool did not fail. There was nothing to record.
Whose failure is this?
Agents sitting in your status meetings is a stretch of the imagination today. The timeline of when this becomes a reality is not the point. Every failure in that scene is already happening in your organization, without a single agent in the room.
And yes, that meeting is a failure of the person running it. She showed up unprepared, she has no agenda, and the meeting decided nothing. She has done that four weeks running and her team has noticed. I am not going to pretend otherwise, and neither will anyone who sits in that room.
But ask what put her there.
We promote people because they are excellent at the work, and then we hand them a job that has almost nothing to do with the work. Planning a piece of delivery and holding a plan together. Defining what a good outcome looks like before anyone starts building. Shaping a solution. Running a meeting that reaches a decision. Preparing materials that let other people decide. Asking the question in the room that nobody wants to ask. Reading what a peer is not saying. Telling an executive something they do not want to hear, early enough for it to matter.
Nobody teaches any of that. We do not coach it, we do not model it, and we do not give people a calendar with enough room to practice it. Then when the results disappoint us, we buy them a tool course.
That is what must change. Not the manager, the way we prepare her.
That is what must change. Not the manager, the way we prepare her.
What is it costing?
Take the meeting in this article. Ten people, one hour, nothing decided.
Put every one of them at a $100,000 salary. That is roughly $48 an hour each, so the meeting costs abut $480 every time you hold it. Run it weekly and you are at $6,250 a quarter and $25,000 a year, for one meeting, to arrive back where you started. Add benefits and overhead and the real number is closer to #32,500.
Now count how many meetings like that sit on your organization’s calendar. Forty of them puts you over a million dollars a year.
Then add what the meetings did not produce. The decisions made twice. The rework. The initiative that sat still for three weeks because nobody set a direction.
Almost nobody counts this. It never appears in a business case and there is no line for it in the budget. It comes out of your delivery dates instead.
What works instead?
Start with how the work is actually happening. Not a survey. A real look at where the week is breaking down, which meetings decide nothing, and which handoffs are getting repaired quietly downstream.
Then build the leader and team’s development plan on what you find and integrate it into their schedule along with performance support and coaching to build the skills.
Then bring in the AI training, into a day where the schedule has been cleared to hold it.
The return on this compounds. A manager who can plan work, hand it off clearly and close a decision gets more our of every tool you buy, every person you hire. and eventually every agent you deploy.
So, look at your AI training budget this week and ask what it assumes about the managers receiving it.
You have funded this pattern once already. How has that worked out for you?
3 Questions You Should Ask
Our AI training assumes managers can define work clearly enough for a tool to help. Have we tested that assumption?
What professional development did we give these people when we moved them from doing the work to directing it? Are they effectively delegating, holding accountability, and driving outcomes?
Pick one recurring meeting on your calendar. What decision does it exist to make, and when did it last make one?