The Hire Nobody's Managing

Most organizations deploy AI agents the same way they used to add contractors — fast, informal, and with almost no accountability structure. Someone in tech identifies the use case, the agent gets deployed, and the first time something goes wrong, the room goes quiet. Nobody owns it.

This episode is about what the CHRO's role actually is in the agentic era. Jackson names three structural traps killing AI governance right now — and three concrete plays to claim the ground before an incident forces you to respond reactively.

What You'll Learn

• AI agents need the same performance architecture as any hire: mandate, output standards, review cadence, and a retirement trigger.
• The three structural traps: treating agents as IT deployments, skipping the performance conversation, and waiting for an incident to build governance.
• Why the CHRO — not IT, legal, or finance — is the only role holding the full accountability picture.
• The exact definition to put on the table at the executive level: any autonomous system that affects business outcomes belongs under workforce governance.
• Three plays to act on now: define the AI workforce, build a parallel performance standard, and get into the AI strategy conversation before decisions are made without you.

Key Quotes

• "If it takes action, produces output, or makes decisions that affect business outcomes — it belongs under workforce governance, not just technology governance."
• "Skipping the performance standard is a choice to let drift accumulate until an incident makes the cost visible."
• "The conversation starts whenever you decide to have it. I'd suggest maybe this week."

Sources for Statistics Cited

• More than half of talent leaders plan to add autonomous AI agents this year — Mercer Global Talent Trends 2026
• Over 80% of business leaders already use AI agents to expand workforce capacity — Mercer Global Talent Trends 2026
• 1.3 billion AI agents projected globally by 2028 — IDC/Microsoft via IT Pro
  
  

Keywords: CHRO leadership, AI agents workforce, AI governance HR, talent architecture, human capital strategy, CHRO altitude, agentic AI accountability, AI performance management, workforce AI deployment, enterprise AI governance

Transcript

More than half of talent leaders plan to add autonomous AI agents to their teams this year. And almost none of them have answered the most basic question you would ask about any new hire: who's responsible for this thing's performance?

Hey there, senior leader, and welcome to the Talent Sherpa Podcast, where senior leaders come to rethink how human capital really works. I'm your host, Jackson Lynch, and today we're going to be talking about something that is already happening inside of your organization, whether you've formalized it or not. AI agents on your team, in your workflows, touching your decisions. And the question no one's really asking is the one that matters most. Who owns the performance conversation with a machine?

Here's what's really happening in organizations right now. They are adding AI agents to workflows the same way they used to add contractors — fast, informal, and with almost no accountability structure. Someone in technology identifies a use case, the agent gets deployed, starts producing output, and the first time someone asks, "Who's responsible if this thing goes wrong?" the room gets very, very quiet, very, very fast.

Now, if you've ever watched a capability get deployed into your organization without a clear owner — and we all have — and then watched it generate noise, risk, and confusion for the next 18 months, this episode names exactly why that happened. And it gives you the framework to make sure that it doesn't happen again with AI.

By the end, you're going to have a clear lens for what the CHRO's role actually is in the agentic era. And it's not what most HR leaders think it is.

Before we get started this week, I want to tell you a little bit about the CHRO Chronicles. It's a weekly paid newsletter written specifically for sitting CHROs. Every week we dive deep into one piece of thinking — mandate clarity, board relationships, the structural forces that shape what is possible at the top of the CHRO house. The community is readers who are exactly who you'd want thinking right beside you. It's only $30 a month. You can find out more at mytalentsherpa.com.

All right, let's get into it.

Let me first describe what's happening in most SMB organizations right now. Somebody on the technology team or in a business unit identifies a task that an AI agent can handle — and those use cases are becoming more and more frequent every time the models get an update. Might be customer inquiries, research synthesis. We might even give agents spreadsheets to analyze, workflow automation, scheduling and coordination. The agent gets deployed and it runs, and the output starts flowing. The agent's producing results. Everyone is feeling good about it. But no one quite owns it. Technology may have built it, operations might use it, finance is watching the cost of tokens, but HR is not necessarily in the room at all.

And that's not hypothetical. The 2026 Mercer Global Talent Trends Report found that more than half of talent leaders plan to add autonomous AI agents to their teams this year. Over 80% of business leaders are already using AI agents to expand workforce capacity. Microsoft has projected 1.3 billion agents scaled globally in 2028.

Even here at Talent Sherpa — your humble correspondent — I figured out how to use Claude Cowork. And I've tried to put all the security things in place, but let me tell you, just doing my base administration, it saves me two or three hours a day. I just went through an entire re-tagging of all of my posts on Substack. I said, "I want you to go pull them. I want you to tell me what the right tags are, and I want you to then erase the ones that are there, reorganize it, and put it back." That would have taken me a day. It took about five minutes. And the work product was better than what I would have done.

So what I'm trying to explain here is whether you're a big company or a small company, the scale is real. The speed is real. And the accountability gap is enormous.

Because here's what happens when you deploy a workforce at that scale without the talent architecture to support it. You get exactly what you get when you deploy a human worker without the talent architecture to support them. Unclear mandates, inconsistent output, no feedback loops. And when something goes wrong, you find out that nobody is actually responsible, and everyone does that human shrug-emoji motion that you've seen.

The difference is that you expect this problem with humans. You have a process for onboarding, performance management, and role clarity — because you have learned the hard way what happens when those things are absent. You have not yet learned those lessons with AI agents. But you're going to. Right now, that tuition is being collected.

The organizations that are getting this right are treating AI agents the way they treat any other workforce capacity. Moderna, for example, merged its technology and HR functions specifically to make clear that AI agents are workforce, not infrastructure. That's a mandate decision. It's an organizational design decision. And it started with the Chief HR Officer and the Chief Technology Officer agreeing on who owns what. Most organizations have not had that conversation yet.

Now, there are three structural traps that are reproducing this problem across SMB enterprises right now.

The first trap is treating AI agents as a technology deployment. When IT owns the agent deployment, the accountability framework applied is an engineering framework. Is it working? Is it secure? Is it compliant? Those are engineering questions. The talent questions are different. What's this agent supposed to produce? How do we know if it's doing that well? Who reviews the output? Who can retrain it? And when does it get retired?

Stephanie Birnbaum was on with us a few months ago, and we talked about the orchestration that no one has ever designed. And that's exactly what we're seeing play out in real life. Because when an organization deploys AI agents and nobody asks those questions, the result is a governance failure. And it's HR's job to notice it and to name it.

The second trap is assuming the performance conversation doesn't apply to machines. I disagree. It totally does. Not in the same form — maybe you can't have the hard conversation with a machine — but the underlying logic still holds. Every performance system needs to be able to answer one question: is the resource producing what we need? And if not, why not?

AI agents need that answer too. They need onboarding standards so you know what correct performance looks like from day one. They need ongoing review so drift gets caught before it becomes damage. They need a retraining process when conditions change — and conditions always change. Organizations that skip this step are going to find out which is way more expensive: building the standard before deployment, or explaining the failure afterwards.

The third trap is waiting for the incident. Most organizations will build their AI agent governance framework after something goes wrong. Maybe not intentionally, but that's what's happening. You might have a compliance violation, a customer interaction that creates liability, an output that was confidently wrong in a high-stakes decision. The incident creates urgency, and then the governance framework gets built reactively — under pressure, probably with someone's job on the line.

And that's the most expensive way to learn a lesson that is literally available right now. It's just a conversation about how to plan for this, and we could be doing it today.

So here's the shift that changes how you read this moment. The question about who owns AI agents is a talent architecture question. And that means the CHRO has both the standing and the obligation to answer it. That requires, candidly, a high-altitude thinker in the role. If you're managing the function, it's not going to cut it.

Think about what a talent architecture actually does. It defines how work gets structured. It creates accountability for outputs. It builds the feedback loops that let an organization learn what's working and what's not. It governs the lifecycle of workforce capacity from entry to exit. Those functions exist for humans. They apply equally to AI agents. The form might look different. The structural logic is identical.

And the CHRO who sees this clearly has a real opportunity. The alternative produces a fractured governance structure with no single lens that covers the full accountability picture. If HR doesn't own the performance architecture for AI agents, somebody else will. It's probably going to be technology. They'll apply that engineering lens we talked about that misses the accountability dimensions. Legal will own it and apply a risk lens that trades capability for compliance. Finance will own it and apply a cost lens that misses the strategic picture. Each of these lenses covers part of the problem. The talent lens is what ties them all together. And the CHRO is the person in the room who should be holding it.

This also reframes what it means to operate at enterprise altitude in 2026. The CHRO who's working on employee experience surveys and headcount planning is operating at a functional altitude. The CHRO leading the discussion when the organization decides how to govern its AI workforce is operating at enterprise altitude. Those are not the same conversation. They do not happen in the same rooms.

So here are three plays that let you claim this ground before the incident forces you to.

Play one: define what the workforce means at the enterprise level and establish that AI agents qualify. What most CHROs are doing is letting the technology team own that definition of "AI agent" by default. They stay in their lane. They work on the human workforce. And the agent fleet grows outside of their orbit.

Don't do that. What to do instead is put a definition on the table at the executive level: an AI agent is any autonomous system that takes action, produces output, or makes decisions that affect business outcomes. If it does those things, it belongs under workforce governance, not just technology governance. And I would argue it's important to think about the broader workforce umbrella as humans, agents, and interim or gig or fractional. The old human nine-to-five workforce model is no longer the full picture. If we want to own the future of that, we've got to pull all of it into one.

Getting that definition agreed to at the C-suite level is hard. And it's the first structural move. It establishes jurisdiction before there's conflict over it. You've got to move fast. You can't govern what you haven't defined. Definition is the entry point to reliability.

Play two: build a parallel onboarding and performance standard for AI agents. When Stephanie was on with us, she said, "I think we might need to have job descriptions for AI agents." And Scott and I readily agreed. By the way, Scott Morris runs AI agents inside his company, and they all have performance reviews.

What most organizations are doing is deploying agents with a technical validation process: Does it run? Does it produce the output? Is it compliant with data standards? Those checks cover whether the agent is working. They leave open whether it produces the right thing, for the right people, reviewed by anyone.

What you need to do instead is extend your performance framework to cover the agent lifecycle. What is the mandate for this agent? What does strong performance look like in measurable terms? Who reviews output and on what cadence? What triggers a retraining or a retirement decision — without the gold watch, by the way?

If you build a one-page standard that answers all those questions for every agent deployment, you are 90% there. It does not have to be complex. It does, however, have to exist. Skipping the performance standard is a choice to let drift accumulate until an incident makes the cost visible.

Play three: claim a voice in the AI governance debate now. Here's the pattern playing out across organizations right now. AI strategy conversations are happening in technology leadership meetings. The CHRO is getting a briefing afterwards. And HR then gets handed the change management piece once the decisions have already been made. That's not the first thing that has ever followed that path.

The move is to walk into the conversation before the decisions are made. And you've got to arrive with a position, not a question. Here's the position: any AI deployment that affects how work gets done, how decisions get made, or how accountability gets assigned is a talent architecture decision. And that requires CHRO input at the design stage.

Bring that position to your CEO and your CTO before the next major agent deployment. Get their agreement in advance. They may not readily agree — and that's exactly why you need to do it before you're deploying. Because that agreement, once it's in place, is the governance mandate you need to do the work.

The structural reason this matters is that organizations that handle agentic AI well are the ones that resolve ownership before it becomes a crisis. Mandate clarity precedes effectiveness. We've talked about that every week. That principle applies to AI governance the same way it applies to every other senior leadership role.

Now, I want to acknowledge that some of you are probably listening to this while your organization has already deployed 14 AI agents and no one told you. That's fine. The conversation starts whenever you decide to have it. I'd suggest maybe this week.

If there's one thing I want you to carry out of this episode, it's this: AI agents are a workforce. The performance architecture that governs people applies equally to machines. And the CHRO who establishes that standard now is doing the most important organizational design work of this decade.

Thank you for spending time with me today. I totally appreciate you being a part of this community of senior leaders who want to rethink how human capital really works. A quick shout out this week to Christine from Denver — thank you for listening, whether you are in Nashville, Tennessee, or Singapore. Our community is growing. We're almost at 50 countries internationally now. And that's because of you.

If you're thinking about how to apply this in your own situation, let me point you to a couple of resources. If role clarity is where you want to begin — and I recommend that it is — reach out to getpropulsion.ai. That's Scott Morris's company. They have AI teammates that enable your leadership to focus on the work that actually drives business outcomes.

And if you're a first-time CHRO, you're preparing to step into the role, or you're going through a major HR transformation, I'd love to work with you. We have built practical tools to help you make an impact from day one. You can find everything you need at the recently rebuilt mytalentsherpa.com. And if you want to listen or read what I write every week, you can do that at talentsherpa.substack.com.

Thank you so much for being here, and for putting up with my voice fighting the Texas allergy season. Until next time — keep raising the bar. Keep asking the accountability questions that no one else is asking. And keep on climbing.