Your New Direct Reports Don’t Sleep: Leading People and AI Agents

It’s Monday morning. Overnight, an AI agent pulled the quarter’s data and built a revenue forecast. An analyst dropped it into a deck before the leadership meeting. A vice president presented it to the executive team with confidence.

By Wednesday, someone notices the number is off. Not wildly. Enough to matter.

So whose call was it? The agent’s? The analyst who trusted it? The vice president who presented it?

Most leaders know the answer: it belongs to the person who put their name on it. What has changed is how many decisions now pass through work that no person fully built, and how many organizations are still working out who owns it.

What has changed for leaders?

AI agents have moved out of the pilot stage and into daily work. They no longer just answer a question when asked. They take a task, work through the steps and hand back a finished product: an analysis, a draft, a forecast, a recommendation.

Companies are scrambling to set rules for it. In KPMG’s Q3 2026 survey of 314 senior U.S. business leaders at companies with $1 billion or more in revenue, half said they are establishing clear protocols for when and how to use AI agents, up from 23% in 2025.

KPMG, Q3 2026 AI Quarterly Pulse Survey; see sources below

That’s real progress. It also means the other half is still working it out, and in the meantime the call falls to individual leaders.

If your team uses agents, you are already leading a mixed team of people and software, whether or not your org chart shows it.

Who owns a decision an AI agent shaped?

You do. That hasn’t changed, and it won’t.

“The model said so” is quickly becoming the new “I was just following the process.” It feels like an explanation. It isn’t one that holds up when the stakes are real. Boards, customers and employees don’t hold software accountable. They hold people accountable.

What has changed is how ownership gets lost. When a person does the work, you can ask how they got there. When an agent does the work and a person passes it along, everyone can assume someone else checked it. The gap opens quietly, and usually you only see it once something goes wrong.

That’s why ownership has to be named before the work starts, not sorted out afterward. The leaders handling this well can say plainly which decisions are theirs, which they’ve delegated to a person, and where an agent’s output is a starting point rather than an answer.

What can’t you hand to an agent?

Agents are getting very good at the parts of work that follow a pattern. What they can’t do is the part of leadership that doesn’t:

• Judgment when the picture isn’t clear. An agent can lay out the options. It can’t own the call when the data runs out and the decision still has to be made.
• Tradeoffs people have to live with. Choosing what to stop, who gets the resources and what waits is a human decision because humans carry the consequences.
• Trust. People follow leaders they believe will be straight with them. An agent can produce the message, but it can’t earn the credibility behind it.
• Making it safe to say what isn’t working. The most valuable information in any organization is the problem people are afraid to raise. That takes a leader, not a tool.

More companies are drawing these lines formally. In the same KPMG survey, 49% of leaders said they are identifying high-risk uses where agents aren’t allowed to make decisions on their own, up from 28% in 2025. That’s movement. It also means half haven’t drawn those lines yet.

At Sagebrush, we call the value a person brings that AI can’t replicate their Human Signal. For leaders, it’s no longer a soft skill on the side. As agents take on more of the work, it is becoming the job.

What do your people need from you now?

When agents arrive, people on your team start asking questions they may not say out loud. What is my work worth now? How will I be judged? Am I training my replacement?

Those are fair questions, and silence makes them louder.

Your managers are feeling the pressure too. Worldwide, manager engagement fell from 31% in 2022 to 22% in 2025, according to Gallup’s State of the Global Workplace 2026 report. The people you count on to lead teams through this change are, on average, running on less than they were a few years ago.

What people need from you is less checking and more translating:

• Why the team is using agents, in terms of the work, not the technology.
• What stays human, said plainly enough that people believe it.
• How performance will be judged when some of the output comes from a tool.

One caution: don’t promise what you don’t control. “No one here will lose their job to AI” is the reassurance leaders reach for most, and if it turns out wrong, people stop believing everything else you say. Say what you know, what you don’t know yet, and when you expect to know more. That honesty holds up better than comfort does.

None of this needs a big announcement. It needs consistency: the same message in the team meeting, the one-on-one and the hallway.

What should you settle this quarter?

You don’t need a finished policy to lead well through this. You do need clear answers to a few questions, even if they’re only answers for your own team:

• Which decisions stay with people, no matter how good the agent gets?
• Who reviews what an agent produces before it reaches a customer, a board or a decision that’s hard to reverse?
• How will you talk with your team about what changes and what doesn’t?
• What are you modeling? Your team will watch how you use agents yourself, including whether you own the output or hide behind it.

If any of these are hard to answer, that isn’t a failure. It’s a sign you’re looking at the right problem. Thinking it through with a second set of eyes is often what turns a hard question into a clear decision.

Will AI make leadership smaller?

There is a quiet fear that AI will make leadership smaller: fewer people to lead, less to decide, more to automate. We think the opposite is more likely. When agents handle more of the routine work, what’s left is the part that was always the hardest: judgment, trust, tradeoffs and the courage to own the call.

That’s also where the pressure lands. Among CEOs, 72% say it is increasingly difficult to know which disruptive forces to prioritize, up from 67% the year before, according to the 2026 AlixPartners Disruption Index. The challenge for most leaders isn’t a lack of information. It’s a lack of clarity.

Your agents will keep getting faster. The question is whether your leadership keeps getting clearer.

If you are the one wondering what AI means for your own role, see Feeling Stuck Mid-Career? What It Means in the Age of AI.