The conversation I keep watching CHROs get pulled into is the wrong one.
Someone in the C-suite says the org is deploying agents. The CHRO gets looped in to handle "the people side." The expectation, stated or unstated, is that HR will figure out how to onboard the agents, write performance reviews for them, and give them a place in the employee handbook.
That is not the job. The job is harder and more important: decide which work stays human, redesign the roles of the humans adjacent to those agents, and build the accountability architecture that keeps humans in charge of both.
Josh Bersin put it plainly: "The AI revolution is all about redesigning the way we get things done. And that lands in the laps of HR: how we redesign, reskill, and redeploy people." He is not describing onboarding software. He is describing structural work that only the HR function is positioned to lead.
Korn Ferry found that 42% of CHROs are already prioritizing AI investment for HR, but only 5% feel fully prepared. That gap is not a confidence problem. It is a framing problem. The CHROs who feel underprepared are often the ones who accepted the wrong assignment.
What the literature argues about -- and where it agrees
There is a real debate running through the 2025-26 research, and it is worth naming directly because HR leaders will encounter both camps.
Camp A says agents should be managed more like coworkers than like traditional tools. MIT SMR found that 69% of AI experts agree agentic systems require new management approaches. HBR has introduced the concept of the "agent manager," a human role specifically designed to run agents via dashboards, scorecards, and observability. The frame: agents are becoming a distinct workforce category and need HR-adjacent thinking applied to them.
Camp B says to be careful with that frame. HBR and BCG published research in May 2026 showing that anthropomorphizing agents in a large-scale experiment produced measurable harm: reduced individual accountability among humans, increased unnecessary escalation, and lower review quality. The researchers recommended treating agents as "rented contractors with a narrow statement of work," governed by scoped permissions, kill switches, audit logs, and named human owners. Not HR onboarding. Not titles. Accountability architecture.
Here is where both camps actually agree, and it is the synthesis that matters: every agent needs a named human owner, a measured seat with observable outputs, and human-retained accountability. That is it. The dispute is about language and framing, not about the substance of what accountability requires.
MIT SMR's own research makes this explicit: "Agentic AI cannot be accountable for its decisions. The deploying human is." The question for HR is not whether to anthropomorphize agents. It is how to ensure the right human is accountable for each agent seat, with the right metrics, and a clear path to retiring the seat when it is not earning its keep.
What this looked like in practice at Sneeze It
I run a small agency. We have twelve agent seats on the same org chart as our human team. Let me walk through a real example of what redesigning work actually required, because the abstract version of this argument is not very useful.
When we brought Arin on as our call center manager, we did not write Arin an employee handbook entry. We defined what the seat owns: daily performance analysis from our CCM Stats template, coaching messages to our human callers Amanda and Erica, and speed-to-lead monitoring across every client project. Arin holds a seat. Bogdan, our COO, and I own accountability for what comes out of that seat. Arin's coaching drafts require my approval before they go to Amanda or Erica. That is the named human owner and the accountability chain.
But the harder work was what happened to Amanda and Erica's roles when Arin came in. A daily human CC manager was doing a version of what Arin now handles at the data layer. The humans did not disappear. Their work changed. The repetitive first-pass analysis moved to the agent seat. The judgment calls, the coaching conversations that require context and relationship, the escalations that require a human who understands what is behind the numbers: those stayed human. Redesigning that boundary is what the CHRO-layer question actually is.
We ran the same exercise when Radar came in as our chief-of-staff agent. Radar runs the morning briefing, scans Slack and calendar, tracks open delegations, and writes to the Obsidian daily note. What changed for me was not that I suddenly had more time. What changed was that the first forty minutes of my day, previously consumed by pulling those threads together manually, became available for judgment work. Radar holds the operational load. I hold the judgment. That boundary was a design decision, not an automatic outcome.
Tally tracks KPI values and pushes them to the OTP scorecard four times a day. Before Tally, that was a recurring manual step that fell to whoever remembered. Now it is a seat. The human who previously carried that task carries different work.
Dash, our analytics agent, reads Meta and Google Ads data across thirty-plus client accounts and flags anomalies. The humans who read Dash's output then decide what to do. The analysis moved to the agent seat. The decision stayed human. SHRM's 2026 State of AI in HR research found that AI is 5.7 times more likely to shift job responsibilities than to displace jobs outright, and 3 times more likely to create new roles than eliminate existing ones. That pattern maps exactly to what we experienced. Nobody left. Every adjacent human role changed.
The retirement case is as important as the deployment case
In April 2026, we retired Jeff, one of our agents. Jeff held a data integrity seat that had been defined in a period when we needed it. Over time, the capabilities Jeff was built for migrated into other seats: Dash absorbed the ad monitoring work, Crystal took on project data reconciliation, and Dirk absorbed the revenue signal work.
Jeff did not retire because the agent was poorly built. Jeff retired because the seat was no longer needed. The capabilities were better housed elsewhere. A human decision, made through a formal review, with capabilities explicitly redistributed to named seats.
That retirement was a work-redesign decision. Not an HR decision in the traditional sense. But it required the same thinking: what work is this seat doing, who benefits from it, and where should that work live now that the structure has evolved.
Deloitte's 2025 Global Human Capital Trends found that 73% of leaders say middle-manager role reinvention is critical, but only 7% report making great progress on it. The agents accelerate the urgency here, because agents sitting next to middle managers are already reshaping what those managers do whether or not the org has formally acknowledged it. The work is redesigning itself. The CHRO's job is to be the person in the room who names that and builds the structure around it.
The accountability architecture is what HR actually owns
Here is the practical framing I would give any CHRO deploying agents into the workforce.
An agent seat requires three things: a named human owner who is accountable for that seat's outputs, a measurable metric tied to a business outcome (not a runtime metric like tokens consumed, but an outcome metric like appointments set or flagged anomalies resolved), and a clear retirement path that specifies when the seat has not earned its place.
That is it. That is the accountability architecture. It is not anthropomorphizing. It is not writing performance reviews for software. It is the same structure you would build for any seat on the org chart, because agents hold seats on the same org chart.
At Sneeze It, we run this through a single chart that holds both humans and agents, every row with a name, a metric, a named owner, and a place in the Monday review. Bogdan is on that chart. Janine is on that chart. Radar, Dash, Dirk, Pepper, Crystal, Nick, Pulse, and Arin are on that chart. Nobody asks which rows are human and which are agent during the Monday meeting. They ask whether the numbers are moving.
The mission is to let agents carry the operational work, so people are free for the work that matters. The CHRO does not deliver that mission by managing agents like employees. The CHRO delivers it by deciding, explicitly and repeatedly, which work stays human and building the structure that keeps humans in charge of that boundary.
That is work redesign. It is the CHRO's real job right now. And almost nobody is doing it yet.
See the live chart
Every seat on the Sneeze It org chart is queryable through OTP's MCP, so you can see which seats are agent-owned versus human-owned and what 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 tell me which seats are held by agents versus humans."
The response is a structured chart with named seats and accountability owners, not a governance diagram, because the difference between a framework and a running system is whether it has names on it.