Rehired at a premium

Gartner put a number this week on something operators have suspected for a while: by 2029, 30% of employees laid off because AI replaced them will need to be rehired — "often at a significantly higher cost." Read that slowly. Three in ten of the cuts justified by AI come back as a job posting. Same seat, more money, and a gap in the middle where nobody knew how the work actually worked.

The line from the same release that will get quoted everywhere: executives will look back and realize "their greatest mistake was believing that work automation was the point, when workforce amplification was the opportunity." I've been saying amplify for a year and it's a little strange to see it in a Gartner press release. But the quote isn't the news. The 30% is.

What actually got cut

Here's how the layoff math gets done. A model handles 70% of a task, so you cut 70% of the people doing it. Sound in a spreadsheet. Wrong in an operation. The person doing the task also carried the 30% the model can't see: which customer's "urgent" is actually urgent, which exception is a one-off and which one is a pattern, why step four exists at all. None of that lives in a system. It lives in the person — and the cut sends it out the door with them.

Gartner's own third shift says it politely: as AI gets embedded in more processes, "context, human judgement and institutional knowledge could be lost." The operator's version is blunter. You didn't fire a task. You fired the map.

Why the rehire costs more

Three reasons, and none of them are about salary bands.

First, the knowledge doesn't come back with the person. A rehire after a year is a new hire with a familiar name. You're paying for training either way; now you're also paying for the year of errors in between.

Second, the reason you're hiring is the reason they left. The system started producing mistakes nobody caught, because the person who used to catch them was the person you cut. Gartner calls this preserving accountability. I call it the human gate: somebody has to own the decision when the machine is confident and wrong, and that somebody was on line 14 of the reduction list.

Third, you're buying back into a tight market. Labor force growth is flat or falling in most of the world, and everyone who made the same cut is now bidding for the same people. The premium isn't a surprise. It's the price of having sold at the bottom.

There's a second number in the release that belongs next to the first: by 2027, 75% of organizations that bank AI productivity gains as cost savings will be eclipsed by competitors that reinvest those gains into modernization and upskilling. Cut and pocket it, and you lose twice — once when the knowledge leaves, again when the competitor who kept theirs pulls ahead.

What we did instead

In transportation and logistics the work is full of residue — the re-typing, the chasing, the reconstructing of yesterday every morning. That's what we pointed the intelligence at. Not the seat, the residue. The person stayed as the gate on anything that left the building. Inbound reaction time went from three-plus days to about a minute, and escalations dropped from seven to ten a day to three to five in six weeks. Nobody was cut to get those numbers. The same people got them, with the room to do it, and the reclaimed capacity showed up as roughly 20% more efficiency — from a team that was already good at the job.

I'm not going to pretend headcount never changes. Roles move. But there's an order of operations, and Gartner just priced what happens when you skip it. Run the system with the human in the seat for a measured quarter first. Then you know exactly which judgment the machine actually replaced — not the one the vendor deck said it would.

Before any AI-justified cut

Write three lines for the seat. What the machine now does. What the person still catches. Who catches it once they're gone. If the third line is blank, that's your 30% — and you can find it now for the cost of a pen, or in 2029 for the cost of a recruiter.

A layoff is a bet that the machine holds the whole job. Gartner says three in ten of those bets come back as a job posting, at a premium. Amplify first. Measure. Then decide what the job really is.

Sources: Gartner, "Gartner Identifies 4 Shifts Shaping the Future of Work" (press release, September 9, 2026)