Most reskilling conversations start in the wrong place.
They start with a list of AI skills. Prompt engineering. AI literacy. Working with autonomous systems. These are fine additions. But they are not the reskilling question that will determine whether your people thrive in an agent-era organization. The real question comes one layer deeper: once the agents are doing the operational work, what exactly are your people supposed to be doing instead?
That is the question the CHRO has to answer. Not the CIO. Not the head of L&D with a new course catalog. The CHRO, because the answer lives inside the work itself, and redesigning work is now HR's mandate.
The gap the data actually describes
Korn Ferry surveyed 15,000 employees across 15 markets in 2025. Forty-eight percent said they fear their job will be replaced by AI within three years. Among tech workers, that number reaches 59 percent. At the same time, 70 percent of senior leaders say their organization has an AI strategy, but only 39 percent of employees agree. That is not a communications gap. That is a work redesign gap. The leaders have a plan for the agents. They do not have a plan for the people.
Josh Bersin put a useful number on the investment asymmetry: for each dollar spent on machine learning technology, organizations may need to spend nine dollars on intangible human capital to make the technology pay off. The technology is the easy part. The people side is where the work is.
SHRM's 2026 State of AI in HR report found that AI is 5.7 times more likely to shift job responsibilities than to displace jobs outright, and three times more likely to create new roles than eliminate them. Read that carefully. The threat to jobs is smaller than employees fear. The disruption to what those jobs contain is larger than most leaders have planned for.
That disruption is where the CHRO's reskilling mandate actually lives.
What happened at Sneeze It
I run a marketing agency. Over the past two years, we built an agent workforce that now holds named seats on our org chart alongside our human team.
Radar is our chief-of-staff agent. Tally pushes KPI values to our scorecard four times a day. Dash analyzes ad performance across Meta and Google for every client account. Dirk runs sales pipeline. Pepper triages email. Crystal tracks project delivery. Nick prospects cold. Arin manages our call center team's daily performance data. Pulse monitors client health.
Each of those seats was previously filled by a human doing operational work: pulling reports, scanning inboxes, chasing statuses, running numbers. Bogdan, our COO, used to start his day reconstructing the picture from a dozen sources. Now Radar gives it to him in the morning briefing. Janine in accounting used to track receivables manually across client accounts. Tally now pushes those numbers to the scorecard on a schedule.
The agents freed real hours. The question we had to answer was not what AI skills our people needed to learn. It was what we wanted Bogdan to do with the time Radar gave back.
The answer was strategy. Decision-making. Conflict resolution. Client relationship work. The things that require judgment, trust, and presence, which are exactly the things agents cannot do.
That reallocation did not happen automatically. It required someone to see the freed capacity, name it explicitly, and redesign what the seat contained. That is the reskilling work the CHRO has to own.
The accountability split that makes reskilling real
Here is where the CHRO has to be careful, because the literature splits on this question and the split matters.
One school of thought says you should manage agents like coworkers, with onboarding, performance reviews, and structured development cycles. MIT SMR research found that 69 percent of experts agree agentic AI demands new management approaches, and the "agent manager" role has emerged in companies running agents at scale. HBR has called agents an "emerging talent category" that HR and procurement need a playbook for.
Another school says that framing is dangerous. HBR and BCG research published in May 2026 found that when organizations treated AI agents like employees, the results included reduced individual accountability, increased unnecessary escalation, and lower review quality among the humans involved. Their model is closer to a rented contractor with a narrow statement of work, governed by scoped permissions, audit logs, kill switches, and, critically, a named human owner.
Both camps are right about something, and the CHRO who ignores either camp will make an expensive mistake.
What both camps agree on is this: every agent needs a named human owner, a measured seat with observable outputs, and human-retained accountability. The difference is in the framing, not the substance.
At Sneeze It, when Jeff, our data integrity agent, stopped earning his seat, a human made the retirement decision through a formal hearing. Accountability never moved to Jeff. It lived with the human who owned his seat. When we say an agent is "onboarded," we mean it has scoped permissions and a clear metric. When we say an agent is "retired," we mean a human decided the seat was not earning its place.
That is accountability architecture. It is not anthropomorphizing. And it is what makes the reskilling possible, because the humans know exactly what they own, and the agents know exactly what they cover.
The three reskilling moves
Reskilling for the agent era comes down to three specific moves, in order.
First, map what the agents absorbed. Go seat by seat and write down what operational work each agent now handles that a human was handling before. Be specific. Not "administrative tasks" but "daily ad performance reporting across 39 client accounts" or "inbox triage and draft responses for 68 client domains." The specificity matters because vague claims about freed capacity produce vague reallocation.
Second, name what the freed time is for. For each seat, write down explicitly what the human owner is now expected to do with the hours the agent returned. If you cannot name it, the hours will fill with overhead. Bogdan's freed time is for strategic planning and COO-level judgment calls. Janine's freed time is for exceptions, relationships with vendors, and financial decisions that require context no agent has. Name the new work before you build a course.
Third, build toward the named work. Only after you have named the freed capacity does the skills investment make sense. If Bogdan's new work is strategic planning, the reskilling investment is in strategic frameworks, financial modeling, and facilitation. If Janine's new work is relationship management with high-stakes vendors, the investment is in negotiation and communication. The skill list follows the work redesign. It never leads it.
What the CHRO owns that nobody else does
Deloitte's 2025 Global Human Capital Trends research found that 73 percent of leaders say reinventing the middle manager role matters, but only 7 percent report making great progress on it. Managers currently spend about 40 percent of their time on administrative work versus 13 percent on people development. Agents can absorb most of that 40 percent. But nobody shifts the 13 percent upward unless someone explicitly redesigns what the manager role contains.
That is the CHRO's work. Not the CIO, who is focused on the agent architecture. Not the CFO, who is focused on the labor cost implications. The CHRO, because the question of what human work is worth doing once agents absorb the operational layer is fundamentally a question about talent, roles, and the design of work.
Let agents carry the operational work, so people are free for the work that matters. That sentence sounds simple. Executing it requires someone to decide, seat by seat, what "the work that matters" actually is. That decision belongs to HR.
The CHRO who makes those decisions well will build a workforce that gets more effective as the agent layer grows. The CHRO who skips them will have a workforce with great AI tools and no clarity on what to do with the time those tools return.
The reskilling conversation starts with the agents. It ends with the humans. The CHRO owns the distance between those two points.
See the live chart
The OTP chart for Sneeze It is queryable by seat type, so you can see exactly which seats are agent-owned and which are human-owned, along with the metrics each seat is accountable for.
In Claude Desktop or Cursor or any MCP client, add this block:
"otp": {
"command": "npx",
"args": ["-y", "@orgtp/mcp-server"]
}
Restart the client. Then ask: "Use OTP to show me the Sneeze It org chart and identify which seats are held by agents versus humans."
The response shows you exactly which operational seats agents hold and which seats humans own with accountability for those agents. That structure is what makes reskilling possible, because you can see where the work lives before you decide where it should go.
Series: AI-era CHRO. Part 20 of an in-progress series. Previous: What HR does when half the workforce is agents.