You've probably had the thought already: AI can do a lot of your job now. Draft the update. Summarize the meeting. Organize the notes you've been sitting on for three weeks. And once you notice how much it can take off your plate, a different question shows up right behind it — if AI can do this much, what am I actually still for?

Here's the honest answer: more than you'd think, and it's the part of the job that was never about speed in the first place.

The trap is treating every part of management like a task to be optimized. If AI can summarize the meeting and draft the first version, it's tempting to let it also handle the performance review, the one-on-one prep, the "great job, team" message — because it's faster, and honestly, a little less awkward. But some parts of the job aren't valuable because they're hard or time-consuming. They're valuable because they're human. They're the moments where someone figures out whether they trust you, whether you understand them, whether their work actually matters to somebody. Hand those off, and you're not becoming a more efficient manager. You're slowly removing yourself from the job of being a good one.

So here are four things I'd never delegate:

• Deciding what matters. AI can track the tasks, create the tasks, and nag everyone about the tasks — that was never really your job anyway. What AI can't do is decide what's worth doing in the first place: the direction, the priorities, the judgment call. Use it to inform that decision, never to make it.
• Judging your people. Let AI organize the evidence for a performance conversation — the notes, the results, the patterns you might have missed. But who gets the raise, the promotion, the tough feedback — that has to come from you, because you're the one who has to look them in the eye and explain it.
• Having the conversation. Coaching, listening, repairing trust — these can't run through a chatbot with you as the middleman. Paying attention to people is still the job, and no dashboard replaces an actual conversation.
• Creating meaning. AI can draft a perfectly nice "great job, team" message. It can't be the one who noticed. Meaning is knowing your work counts; impact is knowing who's counting on you — and if your team ever finds out every kind word came from a bot, you haven't scaled recognition, you've hollowed it out.

None of this is an argument against using AI. Let it take the meeting you didn't need to be in, the notes, the spreadsheet you've been staring at for hours — that's a great use of it. The question was never how much AI you use. It's what you refuse to hand it. I dig into where exactly to draw that line in this week's video — worth a watch if you've been wondering how much of your job you can actually delegate.

or article or podcast​

Here's some more resources I recommend:

Get A Team To Collaborate Fast: Most leaders assume a new team just needs time to gel. It's a reasonable assumption — and an expensive one. Research on aviation crews and surgical teams points to something more precise: teams move fast together because people know who knows what, and how each other operates — not because of hours logged. In this last weeks' video, I broke down why capable teams stall anyway, and the specific moves that close the gap in days instead of months. Click here to watch.​

Navigating Change as a Team: Change isn't the problem—unmanaged change is. When teams face rapid change without a clear plan, the emotional and logistical strain adds up fast, and performance suffers. In my LinkedIn Learning course, I walk through how to build a shared vision and clear priorities for change, how to reassign roles and responsibilities to support the effort, and how to set communication norms that keep things transparent instead of chaotic. You'll also learn how to gauge your team's capacity and resilience so you can sustain performance for the long haul. If your team is navigating change right now, this one's built for you. Click here to get started.​