Somebody Has to Manage the Bots
Management Skillset: Leaders must learn to manage mixed human-AI teams, focusing on coordination and oversight.
HR Involvement: Many HR professionals are disengaged from AI strategy, leaving governance to legal and compliance teams.
Accountability Ownership: Every AI agent's output needs a human owner to ensure accountability for decisions made.
Override Norms: Establish clear guidelines for when team members can and must override AI decisions to avoid confusion.
Mixed Team Performance: Define performance metrics for teams with both humans and AI to prevent ambiguity in evaluations.
As I scanned through Korn Ferry’s TA Trends survey of 1,600 or so talent leaders, I couldn’t help but ask a question that has become popular on social media: what are we doing?
More than half of talent leaders plan to add autonomous AI agents to their teams in 2026. Only 22% of those same respondents believe their leaders can effectively manage teams that combine humans and AI agents.
Read those two numbers together and the story writes itself. Companies are deploying a new class of worker at scale while the people responsible for managing them admit they don't know how.
Keep Reading—and Keep Leading Smarter
Create a free account to finish this piece and join a community of forward-thinking leaders unlocking tools, playbooks, and insights for thriving in the age of AI.
Have an account? Log In
The infrastructure is moving faster than the management thinking is one way to put it.
"HR vendors are already creating employee records for AI agents. Microsoft is issuing them security IDs. The infrastructure for human-AI teams is being built right now," Bryan Ackermann, Korn Ferry's head of AI strategy and transformation, says in the report.
What's being built is the plumbing. Identity, access, records. What hasn't been built is the operating layer above it, the part that answers the questions a manager actually faces when they log on for the day. Things like:
- Who signs off on the agent's output before it reaches a client?
- When a salesperson thinks the agent's forecast is wrong, can she override it, and does she need to document why?
- If a mixed team misses its number, how much of that review conversation is about the humans?
Korn Ferry's report puts the manager's new reality in similar terms, asking how leaders coordinate tasks between humans and machines, when they override AI decisions, and how they handle conflicts between team members. "It's a completely new management skill set," the report concludes.
Jurgen Apello, author of Human Robot Agent, summed it up in an interview with People Managing People earlier this year.
Leadership, now, is about knowing which questions to ask, which constraints to set, and when to intervene. That’s harder than knowing the answers.
HR is Standing Outside the Room
If managing agents alongside people is a management problem, you'd expect the people function to be leading the response. SHRM's State of AI in HR 2026 report, based on a December 2025 survey of 1,722 HR professionals, shows the opposite.
In 52% of organizations, HR has no direct or even collaborative role in overall AI strategy. Governance and oversight of AI sits primarily with legal and compliance functions (37%). And among organizations already using or piloting AI, 49% have policies regulating how their workforce uses it. Of those that do, only a quarter describe their policies as clear and built to last. More than half say their rules are too restrictive and written for tools that will be obsolete in a year.
The more uncomfortable finding is that HR professionals aren't fighting for the job. Only 16% believe HR should lead change management around AI adoption, and just 15% think HR should lead training employees to use it. For four of the six implementation activities SHRM measured, more HR professionals preferred no involvement at all over leading.
That leaves the rules for human-agent teams being written by lawyers, whose mandate is risk, and IT, whose mandate is uptime. Neither is responsible for whether a team functions. When legal writes the policy, you get an acceptable-use document. Nobody gets a management system.
Four Decisions That Make This Manageable
None of this requires a task force or an eighteen-month framework project. It requires leadership to make four decisions and put names next to them.
- Every agent's output gets a human owner.
Accountability cannot diffuse into software. If an agent drafts the client deliverable, a named person answers for it, the same way a manager answers for a direct report's work. If you can't say who owns an agent's output, that agent shouldn't be in production.
- Set override norms before the first dispute, not after.
People need to know when they can overrule an agent, when they must, and that doing so carries no penalty for slowing things down. A team member who lets a bad agent decision through because overriding felt career-risky is a failure of policy, not of judgment.
- Define what performance means on a mixed team.
If the agent's throughput counts toward the team's output, decide now whose number it is, how it's weighted in reviews, and what the humans are actually being evaluated on. Ambiguity here will surface in your first review cycle, and it will surface as a grievance.
- Assign ownership of the people side of agent policy, explicitly.
If HR won't take it, the COO should, because someone with responsibility for how work actually gets done has to hold it. The SHRM data suggests this decision is being made by default in most organizations. Default is the wrong way to make it.
The deeper issue is that agents are being treated as a procurement decision when they are a management decision.
AI adoption isn’t about tools. It’s about rethinking how work gets done. They’re treating it as a technology problem instead of a leadership problem.
A company would never add fifty contractors to client teams without deciding who supervises them, how their work is checked, and what happens when they get something wrong.
That standard shouldn't drop because the new workers are software. Organizations adding agents this year are conducting a live experiment in mixed-team management, and the leaders running it owe their people the same clarity they'd owe any team taking on new colleagues.
