AI and Employee Engagement Can Go Either Way. Leaders Decide Which.

Maya and Marcus have the same job at two different companies. Senior client managers, both handed Copilot licenses back in February.

By summer, Maya likes her job more than she has in years. Her company opened the rollout by asking her team two questions: which parts of your work drain you, and which parts would you fight to keep? AI took the status reports and meeting recaps. The hours came back to the client work that made her take the job in the first place.

Marcus's company sent a license and a launch email. Six months later, AI writes the first draft of everything, including the data narratives that made him the person leadership listened to. His job became reviewing output. He's quietly wondering what he's there for.

Same tool. Same role. Opposite directions. And the newest research on AI and employee engagement says the difference between these two companies has nothing to do with which AI tool they chose. 

Gallup's July 2026 study found that simply giving employees access to AI neither improves nor harms engagement. The license itself does nothing. What separates Maya's company from Marcus's is everything that happened around the tool.

Where AI and employee engagement come apart

Marcus's company did what most companies do: deploy, announce, hope. Nobody decided AI should hollow out his job. It happened by drift, because nobody decided anything at all.

To see why that drains people, look at what actually drives engagement. Gallup's decades of workplace research keep landing on the same core conditions: people are engaged when their opinion counts, when they get to do what they do best every day, and when they know what's expected of them. A careless AI rollout bruises all three at once. Nobody asked for their input. The work they do best just got automated. And nobody has told them what they're still responsible for. Take those away and you get quiet quitting with excellent attendance.

Meanwhile the dashboard looks great. None of this shows up in usage metrics. Marcus uses Copilot every single day.

The stakes run in both directions, though. Protect people's sense of value and the payoff is measurable: engaged employees stay, put in effort nobody asked for, and take better care of customers. One large field study of customer support agents found AI assistance lifted productivity 15 percent and came with lower turnover, and customers even treated the employees better. Value felt becomes value returned.

Five steps that make AI and employee engagement climb together

Maya's company didn't spend more on technology. They ran a sequence, and it's one any leader can run, starting this week.

Step 1: Run the Two Lists with your team. Before anyone automates anything, get your team in a room and build two columns. Friction: work that's boring, overhead, routine, easy to delegate (we call it BORE work at Modern Workery). Fuel: the judgment, creativity, and relationships people came here for. Doing this together, instead of deciding for them, is the point. McKinsey found employees are three times more ready for AI than their leaders believe, yet nearly half of leaders say they wouldn't involve nontechnical employees in early AI work. Giving people a seat at that table restores the "my opinion counts" driver on day one. Open the meeting with this:

"We're using AI to buy back time from busywork so more of your week goes to the work only you can do. You know your work better than I do, so you're helping decide what goes where."

If you want a rehearsal first, run the exercise solo on your own job. It takes ten minutes and it will change how you facilitate it.

Step 2: Draw the line out loud, especially where AI won't go. From the two lists, name where AI should be used, and more importantly, where it shouldn't. The Fuel column is a commitment, not a suggestion, and saying it plainly answers the question every employee is silently asking about their own value:

"AI drafts. You decide. The client relationships, the recommendations, and the final call stay yours."

Step 3: Let each person pick their first handoff. Ask everyone: "What's one task from the Friction column you'd love to never do again?" Then help them hand that one to AI first. Choosing it themselves protects the autonomy that engagement runs on, and the early win builds trust for everything after.

Step 4: Reinvest the reclaimed hours on purpose. When AI hands someone four hours back and those hours evaporate into more email, the engagement gain evaporates with them. Decide together where recovered time goes: the client conversation, the improvement project, the skill they've wanted to build. And close the loophole people are worried about:

"The hours AI saves you are yours to reinvest. They're not a signal that I expect more volume."

Step 5: Keep it alive in your one-on-ones. Add one question: "Where did AI help you this week, and where did it get in the way?" This is where the engagement math gets interesting. Gallup found manager support was the strongest factor in the whole study: 48 percent engagement when managers actively support AI use, 30 percent when they don't. Stack frequent use, a clear plan, and that manager support together and engagement reaches 53 percent, against a U.S. average of 31. Very few levers in this field move anything 22 points, and this one is mostly conversations.

Try This: The 20-Minute Kickoff Agenda

Steps 1 through 3 fit in a single team meeting. Here's the agenda:

1. Minutes 1 to 5: everyone privately lists their own Friction and Fuel work.
2. Minutes 6 to 12: share and build the team's two columns on a whiteboard.
3. Minutes 13 to 17: you draw the line and say what stays human, out loud.
4. Minutes 18 to 20: each person names the first task they'll hand to AI this week.

Your homework afterward is steps 4 and 5: decide where the reclaimed time goes, and add the question to your one-on-ones.

AI is going to change what your team's week looks like either way. The only question is whether the hours it frees up flow toward the work that makes people want to stay, or whether the meaning quietly leaks out while the dashboards say everything is fine. Maya and Marcus started the same February with the same tool. Which company is yours becoming?