🚩Tactical Memo 091: Your Team Thinks AI Is a Layoff Memo

Welcome to Tactical Memo: Tactical advice for managers running teams, projects, and operations in the AI era. Read by 18,000+ leaders at Amazon, Deloitte, and Google. If someone forwarded this to you, join our free community here.

Quick announcements

In September I'm running 2 FREE workshops.

If you're an AI-era manager responsible for delivering outcomes, these are for you.

September 9th: How AI Can 10x Your Project Delivery Speed (and your career)

I'll walk you through the exact 5-step sequence I use to offload the busywork that steals 20 hours a week.

You 10x your delivery speed. Then you reinvest that time into the judgment calls that actually build your career.

Register free here

September 16th: The New Way To Manage: Humans and AI Agents on One Team

Your team isn't all human anymore.

I'll show you what management looks like now, and exactly how the best managers are running people and AI agents on one team.

Register for free here

Both sessions are capped at 200 seats. And they're filling up fast.

If you plan on coming, grab your spot now (one or both, your call).

Hope to see you there 🔥

The most common AI rollout failure makes NO noise at all 👀

Last year I audited an AI rollout for a mid-sized services firm.

The dashboards looked great. 91% of the team had active licenses.

Then I sat with the team for a week.

They opened the tool when their manager walked by. Every deliverable was still built the old way, then run through AI once so the usage log had an entry 😅

Nobody argued with the mandate. They just waited it out (the way they waited out the last three initiatives).

Leadership read the silence as adoption.

The team read the rollout as a layoff memo with extra steps.

An AI rollout is a TRUST problem before it's a technology problem.

Here's what I do now on every rollout:

1/ Answer the layoff question before anyone asks it

In the first meeting, I say the quiet part myself:

"You're wondering if this tool is here to replace you."

Then I give the honest answer for that specific team. In writing. With my name on it.

Silence on that question doesn't calm people. It confirms their worst guess.

2/ Start with the work everyone hates

I never point AI at anyone's craft first.

I point it at the status reports and the reformatting that eats Thursday afternoons.

One team's weekly client summary took an analyst 4 hours to compile. AI cut it to 40 minutes, and she got her Thursday back 🙏🏼

Nobody defends the work they resent.

3/ Give the pilot to the loudest skeptic

I skip the enthusiast.

On a logistics project, mine was a coordinator with 15 years in who called the tool a toy at kickoff.

Six weeks later she presented her OWN results to the team. In her own words.

Skeptics make the best salespeople (everyone knows they weren't bought).

4/ Kill the surveillance signals

I don't track individual usage. And I say so out loud.

The moment people believe the login data feeds a performance review, the tool becomes evidence. And they start managing the evidence 😩

I look at team-level output instead: cycle time, hours returned to the calendar.

Watch the work, and people use the tool. Watch the people, and they perform for the log.

5/ Show my own bad prompts

Every other week I screen-share my actual AI chats with the team. Including the ones that went nowhere.

They watch me try a prompt, get garbage back, and rework it three times.

That does more for adoption than any training deck.

People won't use a tool in public until someone senior fails with it in public first.

6/ Put what AI will NOT touch in writing

One page. Decisions that stay human: hiring, performance reviews, anything a client sees without review.

The team helps write it, so they trust it.

When the fear spikes again (and it always does), we pull up that page instead of starting the argument over.

A boundary in writing beats a reassurance in a meeting.

7/ Give the saved hours back

Six weeks in, I total the hours the tool saved. And I make sure the team keeps a visible share of them.

If AI saves your people 5 hours a week and every hour gets refilled with more output, you've taught them that efficiency is a trap.

One team banked the savings into no-meeting Friday mornings 🔥

People protect a tool that pays them.

The rollouts that fail are quiet.

The rollouts that work are LOUD. Because people only argue about a thing they believe is staying.

Before your next AI announcement, do one thing: answer the layoff question in plain language, in writing, before anyone asks. Everything else builds on that answer.

P.S. If your adoption dashboard looks perfect and your output hasn't changed, that's not a win. That's the sandbagging.

Until next time,

Justin Bateh, PhD
Founder and Editor
Tactical Memo