The question sounds abstract until you live the operational version.
Who decides when a new agent seat gets created? Who sets the metric the agent is measured on? Who reviews the agent's output when the numbers slip? And when the agent stops earning its place on the chart, who makes the call to retire it?
At most companies running agents today, the honest answer is: nobody knows. The engineering team deployed it. The product team uses it. The exec team approved the budget. But nobody owns the lifecycle of that seat the way a manager owns the lifecycle of a direct report.
That gap is not a technology problem. It is an organizational design problem. And it is one the CHRO is positioned to close.
The lifecycle nobody owns yet
Every agent seat on a working org chart passes through four phases.
It is created when a gap appears on the chart that no current seat covers. It is scoped when someone defines what the seat owns, what it does not own, what it is measured on, and who the named human is accountable for what it produces. It is measured when its output appears on the same scorecard as the human seats around it, reviewed on the same cadence, with the same seriousness. And it is retired when a human decides the seat no longer earns its place.
These four phases describe a workforce lifecycle. HR has owned this lifecycle for human seats for decades. The question is whether HR's mandate extends when the seat is filled by an agent rather than a person.
HBR Analytic Services surveyed 603 business leaders in late 2025 and found that only six percent fully trusted agents with core processes. Only twelve percent had risk and governance controls fully in place. The agents are running. The lifecycle governance is not.
That is the gap.
What the literature says, and where it splits
The research on how to govern agents in an organizational context breaks into two camps. Both are worth reading carefully before you build anything.
Camp A, anchored by MIT SMR and HBR, says agents need to be managed more like coworkers than software tools. Sixty-nine percent of experts MIT SMR surveyed agreed that agentic AI requires fundamentally new management approaches. HBR described a new human role called the "agent manager," someone who governs agents through dashboards, scorecards, and observability layers. The framing is: treat the agent as a member of the team who happens to be non-human.
Camp B, from a Kropp et al. study published in HBR in May 2026 with BCG, ran a large experiment and found that anthropomorphizing agents produced specific and measurable harms: reduced individual accountability, increased unnecessary escalation, lower review quality. Their prescription was explicit. Treat the agent like a rented contractor with a narrow statement of work. Use scoped permissions, kill switches, audit logs, and a named human owner. Do not write performance reviews for the agent. Do not give it a title. Keep accountability where it belongs, which is with a named human.
I have run a hybrid team for eight months. I have read both camps. The synthesis I keep coming back to is this: they are arguing about language more than substance.
Neither camp says agents should run without human oversight. Neither camp says accountability can move to the agent. Both camps converge on the same three requirements for any agent seat. First, a named human owner. Second, a measured output in business-outcome language. Third, human-retained accountability for what the agent produces.
What Camp B is warning against is not governance. It is the cognitive drift that happens when organizations use HR language in ways that obscure who is actually responsible when something goes wrong. "Onboarding" an agent as if it were a new hire creates the impression that the agent now carries its own accountability. It does not. The human who deployed it does. MIT SMR stated this plainly: agentic AI cannot be accountable for its decisions. The deploying human is.
OTP's one-seat-one-owner model is accountability architecture, not anthropomorphism. The seat gets a metric and a named human owner. The human owner answers for what the seat produces.
The lifecycle in practice
At Sneeze It, eleven agent seats currently hold rows on our org chart alongside Bogdan as COO, Janine in accounting, and Kristen in creative. Radar runs our daily operations and morning briefings. Dash analyzes advertising performance across Meta and Google. Dirk owns sales pipeline. Tally pushes KPI values from our local systems to the OTP scorecard. Pepper triages client email. Crystal tracks project delivery. Arin manages the call center team through daily Slack coaching. Nick runs cold prospecting in health and wellness. Pulse monitors client retention risk.
Each of those seats was created because a specific gap appeared on the chart. Each has a single business-outcome metric. Each has a named human owner who is accountable for what it produces. That owner is always me or a member of my leadership team.
And each has a defined exit condition.
In April, we retired Jeff, an agent who had held a data integrity seat on the chart. The retirement was not a software deprecation. It was an organizational decision, reached through a formal hearing where every capability Jeff held was documented, evaluated, and reassigned to a named seat. Radar inherited account-level status monitoring. Dash absorbed budget pacing and reconciliation. Dan, my strategic co-founder seat, absorbed the data architecture and blind spot identification work.
Jeff did not advocate for his own survival. The decision was mine to make. The accountability was never his to hold.
That is what a properly designed lifecycle looks like. The creation was a human decision. The scoping was a human decision. The measurement ran on the same human scorecard. The retirement was a human decision. The agent executed in the middle.
Why the CHRO is the right owner
The Korn Ferry Workforce 2025 survey of 15,000 employees across 15 markets found that 42 percent of CHROs were prioritizing AI investment but only five percent felt fully prepared. The gap is not motivation. It is mandate clarity.
The CHRO's mandate has historically been: design the roles, hire the people, develop the people, manage the exits. The lifecycle. When agents fill roles, the lifecycle does not disappear. It expands. The decisions are the same in structure but different in content.
Designing the role means scoping the agent's permissions before it is deployed, not after. Hiring the agent means selecting a capability set matched to the gap on the chart, not browsing a marketplace because the demo looked good. Developing the agent means updating the brief when the metric slips, changing the SOP when the output drifts, investing in the right training data or tooling when the seat is underperforming. Managing the exit means holding a retirement hearing when the seat stops earning its place.
Bersin framed the HR imperative well: the AI revolution is about redesigning the way work gets done, and that redesign lands in the laps of HR. That is not a metaphor. It is an operational description of what happens when the CHRO takes the agent workforce lifecycle seriously.
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 them. When Radar absorbed the operations function that a human chief of staff would have held, I did not eliminate a human seat. I freed the humans adjacent to that seat to move into the judgment work that agents cannot do. Bogdan focuses on strategic operations. Kristen focuses on creative direction. The humans who might have spent forty percent of their time on the administrative load those agents now carry are instead spending that time on the decisions that compound.
Deloitte's 2025 Global Human Capital Trends report found that managers spend roughly forty percent of their time on administrative work versus thirteen percent on people development. The inversion of that ratio is what the agent workforce makes possible. The CHRO is the person who engineers the inversion.
Let agents carry the operational work, so people are free for the work that matters. The CHRO is the one who decides which is which and builds the architecture that holds.
The three questions that make the mandate concrete
The CHRO who takes ownership of the agent workforce lifecycle needs to answer three questions for every agent seat on the chart.
Who owns this seat? Not a team. Not a department. One named person who answers for what the seat produces, whose performance evaluation is affected by whether the agent is performing, and who makes the call when the agent needs to be redirected or retired.
What is the seat measured on? The metric is a business outcome, not a runtime statistic. It lives on the same scorecard as the human seats around it and is reviewed on the same cadence with the same seriousness. Tally's metric is KPI push accuracy and cadence. Dash's metrics are pattern detection rate and alert precision. Radar's metric is operational coverage across every briefing channel. These sit next to Janine's receivables numbers on the Monday dashboard.
What are the conditions for exit? The exit condition is defined before the seat is filled. The exit decision is a human decision. The agent does not accumulate tenure. The seat either earns its place on the chart or it does not, and the evaluation belongs to the person who owns it.
These are not technology questions. They have always been HR questions. When agents hold seats, HR's mandate extends to cover them.
See the live chart
The Sneeze It org chart, including every current agent seat with its named human owner and business metric, is queryable through the OTP MCP.
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 every agent seat, its named human owner, and the metric it is held to."
The accountability architecture is live. The lifecycle decisions behind each row are what this series is about.