Making Is Cheap Now. Finishing Isn’t.

You’ve probably gotten a message like this recently: “I put this together with AI. Can you take a look before it goes out?”

It looks great. It’s long, well formatted, and confident. And as you start reading, it becomes clear that the person who sent it hasn’t read it either.

This is one of the most common things happening at work right now, and almost nobody is talking about it.

We all got promoted at the same time

AI handed every person in every company something that used to be a privilege of seniority: someone to do the first draft. Research, analysis, decks, summaries, reviews. Work that used to go to an intern or a junior analyst can now be produced by anyone, for anything, in minutes.

That makes everyone a manager. And most people have never been taught how to manage.

Bad management has always looked the same. Work comes back from the intern, and it gets forwarded without being read.

“Did you review it?”

“No, it’s fine.”

That habit existed long before AI. What changed is that everyone now has their own intern, producing more than any intern ever could, all day long. The habit scaled with it.

Where the work went

Here is the shift underneath all of this. Producing something now takes seconds. Checking it still takes as long as it always did. Someone has to read it, test the numbers, question the assumptions, and decide whether it’s right.

So output has exploded, and the capacity to verify it hasn’t moved at all. Every document, analysis, and “here’s what I found” creates work for whoever reads it next. When the person who generated it doesn’t do that work, it doesn’t disappear. It moves down the line, usually to the person with the most accountability and the least time: the founder, the executive, the one whose name ends up on it.

That’s why so many senior people feel busier than ever right now. They aren’t doing more of their own work. They’re finishing everyone else’s.

Busy isn’t the same as done

Most people using AI this way feel very productive. In a sense they are; they’re producing more than ever. But a lot of that production is digging. Run a question, get back a long list of issues, pass it along. Ask for an analysis, get forty pages, forward it.

What gets skipped is the question that comes before all of it: is this worth doing? Will the answer change anything, or am I doing it because it’s easy now?

More information is rarely what a decision needs. Most good decisions come down to a few numbers and a judgment call. Burying the person making that call under more output doesn’t help them make it.

Nobody is deciding

The quietest skip is the decision itself. Output arrives with “here you go” instead of “here’s what I think we should do.”

Before AI, avoiding a decision was harder to hide; eventually you had to show up with an answer. Now you can show up with something thorough and polished and look like you did the work. AI didn’t create people who avoid decisions. It gave them very good cover.

Your name is still on it

There’s a simple test for all of this. If something goes out with your name or your signature on it, you own it. The tool doesn’t. “AI reviewed it” won’t mean much to an investor who finds an error in your deck, or to anyone else relying on what you sent.

Accountability didn’t move when the work did. It stayed with the person who hit send.

The job now is the last 10%

AI is very good at the first 90% of a lot of work. The last 10% is where the value has always been, and now it’s where nearly all of it is:

• Check it: read every line and catch what’s wrong.
• Fix it: correct what’s broken, then look for what isn’t there. What did AI miss? What context did it never have? What did it state with complete confidence that simply isn’t true?
• Finish it: get it to the point where it can actually go out the door.
• Decide, and share the decision: make the call, then tell people in a few lines what you decided and how you got there. Keep the full backup ready for anyone who wants it, but don’t make everyone dig through it to find your answer.

That’s what good managers and good editors have always done. Some people would call it perfectionism. It’s becoming the core of what a person contributes.

If AI can do entry-level work, then doing entry-level work is no longer where your value is. Forwarding AI output isn’t a job. Owning it is.

Be the human

The people who matter most over the next few years won’t be the ones who use AI the most. They’ll be the ones who take what it produces and make it right.

So before you send the next thing, ask the questions a good manager asks. Was this worth doing? Did I read it? What’s missing, and what isn’t true? What did I decide, and can I explain why in a few lines?

Then send it. Be the human in the loop.