AI agents have arrived on HCM platforms, requiring HR leaders to now also govern digital workers.
Key takeaways
Effective hybrid workforce plans require decomposing work by task, not just headcount or titles.
CHROs who define governance, cost allocation and AI agent rights now can avoid sprawl later.
A hybrid workforce—one where artificial intelligence agents and people share the same operating model, cost centers and performance expectations—is no longer a planning scenario. It is happening now as major human capital management (HCM) platforms add digital employee management capabilities. Whether chief human resources officers (CHROs) are ready for the change or not, the platforms HR professionals use to hire, track and manage employees will soon carry AI agents alongside human workers.
As the hybrid workforce becomes a reality, many middle market companies are still finding their footing in using AI strategically. According to the RSM Middle Market AI Survey 2026, 86% of middle market organizations have already integrated AI into their operations, with 97% reporting satisfaction with the delivery of business value. However, only 36% have fully embedded AI across core processes—indicating that while adoption is widespread, enterprise-scale integration remains limited. This gap highlights a critical shift: the challenge is no longer whether organizations are using AI, but whether they can scale its impact beyond early successes.
What is a hybrid workforce of people and AI agents?
A hybrid workforce is a workforce in which human employees and AI agents can take autonomous, multistep actions to complete tasks, operating together within a shared organizational structure. An AI agent, sometimes referred to as a digital worker, is distinct from a simple automation or a chatbot. A chatbot answers a question, while an AI agent can plan a sequence of actions, interact with external systems, make decisions within defined parameters and report outcomes, requiring limited human involvement to direct each step.
A hybrid workforce creates new opportunities for efficiency and productivity. It can sharpen focus on priorities and higher-value-added activities, but HR strategies will need to evolve beyond a traditional, structured model to a more dynamic approach. That is a significant structural shift for any HR function built around job descriptions, headcount targets and salary bands.
What HR must govern that it has never governed before
HR’s new responsibility in managing AI agents is not technical configuration. IT typically manages the technical issues. HR’s role is organizational: determining who owns the agent, what cost center carries it, what decisions it is authorized to make and how its performance is measured and remediated if it falls short.
Today, those questions are typically answered by default, with individual business units spinning up AI agents to automate finance tasks, customer service interactions or payroll processing.
According to the RSM US MMBI Workforce 2026 report, 65% of middle market firms are outsourcing or considering outsourcing IT functions, with 40% considering outsourcing customer service and 38% considering outsourcing payroll. Many of those same functions are candidates for deploying AI agents. If HR is not at the table when those decisions are made, agents will effectively be “hired” without an owner, a governance structure or a cost allocation framework.
Gartner projects that by 2028, 33% of enterprise software applications will include agentic AI and 15% of day-to-day work decisions will be made autonomously by agentic AI systems. As organizations evaluate these technologies, the time to build the governing framework is before deployment, not after.
Governance should extend beyond the AI agent itself. Organizations also need clear expectations for how employees interact with AI agents, how decisions are reviewed, what training is required and how accountability is shared between humans and AI.
Deanna Balkcom, Director, RSM US
How to decompose work between humans, agents and hybrid teams
Task decomposition is the core planning discipline with a hybrid workforce. Rather than asking “Which roles can AI agents replace?” CHROs should ask which tasks within each role are candidates for AI-agent execution, and what that means for the human component of that role.
A practical decomposition framework consists of three categories:
Human-led tasks: This category includes activities that require judgment in ambiguous situations, stakeholder relationships, ethical reasoning or nuanced communication. Activities involving employee relations, ethics, organizational culture and sensitive employment decisions should continue to rely on human judgment, even as AI tools support the process.
Agent-led tasks: These tasks are structured, high-volume, rules-based activities with defined inputs and outputs: benefits enrollment processing, payroll calculations, compliance data aggregation and screening-criteria matching in recruiting. These are strong candidates for deployment agents now, with appropriate oversight.
Hybrid tasks: These are activities in which AI agents can accelerate or support human decision making without replacing human judgment. For example, a hiring manager may review an AI-generated candidate summary, but the manager remains responsible for the hiring decision. Similarly, an HR business partner who relies on AI agent-produced workforce analytics still interprets the findings. Hybrid tasks are the most common in knowledge-intensive middle market organizations.
This decomposition structure shapes how traditional workforce planning looks. Rather than focusing on building a headcount plan, HR leaders build a capacity plan that accounts for human full-time equivalents (FTEs) and AI agents.
As work is redistributed among employees and AI agents, leadership expectations should evolve as well. Managers will increasingly oversee teams composed of employees and AI agents, requiring new skills related to workload allocation, quality oversight, coaching and performance management.
Building cost allocation and governance for digital workers
AI agents involve real costs, including licensing fees, computing costs, integration maintenance, quality oversight and the labor costs of the humans who supervise them. When those costs live in the IT budget by default, HR and finance lose visibility into the true cost of work and cannot make sound decisions about where to invest in human versus AI-agent capacity.
CHROs and chief financial officers should establish a cost-of-work model that treats AI agents as managed capacity, similar to how managed services arrangements are structured. The key decisions involve:
- Cost center ownership: Which business function owns the agent and carries its costs? The function that benefits from the AI agent’s capacity should bear those costs for workforce planning purposes, not IT.
- Budgeting model: AI agent capacity costs behave differently from salary costs. Capacity-related costs scale with volume and can be turned up or down faster than headcount. Build a variable-capacity line in workforce budgets to account for this.
- Performance metrics: What output or business outcome justifies the AI agent’s cost? Define this before deployment the same way you would define performance expectations for a new hire.
- Governance and escalation: Who decides when an agent’s decision should be escalated to a human? Document the decision rights before the agent goes live.
Organizations that deploy AI agents onto existing processes will likely capture limited value. Governance defines the framework, but change management drives adoption. Organizations should communicate how AI will change work, address employee concerns and provide opportunities for employees to build confidence using AI responsibly.
Workforce development and capability building
One of the greatest workforce challenges associated with AI adoption may not be workforce reduction—it may be the loss of the apprenticeship model and other opportunities for employee development. As AI agents perform much of the entry-level work, organizations must answer a difficult question: Where will future experts come from? The employees who challenge a manager's assumptions, identify flaws in an agent's recommendations and eventually become leaders typically develop those capabilities through years of hands-on experience. When organizations automate away those developmental opportunities without creating new learning pathways, they may find themselves with highly efficient processes but too few people capable of exercising independent judgment when those processes fail.
Steps CHROs should take now
The practical sequence to building a hybrid workforce governance framework includes the following steps:
- Conduct a task-level workforce audit. Map the 10–20 most time-intensive activities in each HR-adjacent function against the three-category decomposition framework. Identify where AI-agent deployment is already happening without a formal governance structure.
- Define digital worker ownership. Establish ownership for each active or planned AI agent deployment, including the responsible business leader, the associated cost center and the HR record (even if the HR record is a configuration entry rather than an employee profile).
- Align HCM platform configuration with digital worker tracking. Work with IT and your HCM vendor to understand what digital employee management features are available in your current platform version. Evaluate whether your current configuration supports AI agent tracking.
- Establish a decision-rights framework. Document the boundaries between AI agent-autonomous decisions and human-required decisions for each deployed AI agent. Include an escalation path and a remediation process.
- Revise workforce planning inputs. Update your workforce planning models to include AI agent capacity alongside human FTE projections. Introduce a cost-of-work metric that spans both.
- Build an HR policy for AI agents. Formalize the activities AI agents can and cannot do within your organization, how their outputs are reviewed and how accountability flows when an AI agent produces an error.
- Assess workforce readiness. Evaluate whether leaders and employees have the skills needed to supervise, validate, collaborate with and challenge AI-generated outputs effectively. Technology readiness and workforce readiness should advance together.
As AI agents become part of the operating model, organizations should also revisit how employee performance is measured. Traditional productivity metrics may no longer reflect value creation when AI performs routine work. Performance expectations should increasingly emphasize judgment, collaboration, innovation, problem-solving, relationship management and the effective use of AI.
Deanna Balkcom, Director, RSM US
The takeaway: Build the framework before the platform builds it for you
The hybrid workforce is not a future state that HR needs to plan for. It is being in the platforms HR professionals use today, under the same subscription agreements already in place. RSM’s AI Survey shows that AI is already embedded in most organizations’ operations, but not yet fully integrated across core business processes—meaning hybrid workforce models are forming before governance structures and operating models are fully defined.
The organizations that manage the transition well are those that make key decisions now, before deployments increase, including who owns the AI agent, what work it governs, what it costs and how its outputs are overseen.
CHROs who wait for a fully formed AI strategy from the enterprise before addressing hybrid workforce governance are likely to find AI agents already operating in their organization without clear accountability. The framework does not need to be perfect. It needs to exist.
RSM’s human capital management and managed services practices work closely with middle market companies to define operating models for hybrid workforces, including task decomposition, cost-allocation frameworks, governance structures and HCM platform configuration. Contact our HCM advisory team to learn more about building a digital worker governance framework for your organization.
Frequently asked questions
A hybrid workforce AI agent model is an operating structure in which AI agents work alongside human employees within a shared organizational design. Under this model, the organization’s HR governs human FTEs and digital workers, including defined task boundaries, cost allocation approaches and decision rights for each. The distinction from traditional automation is that AI agents operate with a degree of autonomy: they plan, execute and report outcomes without requiring step-by-step human direction.
Workforce planning in a hybrid environment shifts from role-based headcount models to activity-based capacity models. Rather than projecting how many people you need to fill a given role, you decompose the work within that role into tasks and assign each task to the humans, agents or a hybrid best suited for it. The resulting workforce plan accounts for human FTEs alongside AI agent capacity, with a combined cost-of-work metric that allows finance and HR to make comparable investment decisions.
Responsibility for managing AI agents should be distributed across functions in a structured way. The business unit that benefits from the agent’s work often owns the operational performance and carries the budget. HR may own the governance framework, including task boundaries, decision rights, escalation policies and the organization’s formal AI agent policy. IT generally owns technical configuration, security and system integration. Without a defined RACI matrix (responsible, accountable, consulted and informed), agent deployments default to IT ownership by name and no one’s ownership in practice, which creates accountability gaps when agents produce errors or exceed their intended scope.
Without a governance framework, AI agent deployments can lead to AI agent sprawl, accumulating without clear accountability. Specific risks include: costs that are misclassified or invisible in workforce budgets; agent decisions that lack a defined escalation path when edge cases arise; compliance exposure when agents process regulated data without documented oversight; and workforce confusion about which decisions require human judgment. Governance does not require a fully developed enterprise AI strategy, but it does require a documented owner, defined task scope and a decision-rights policy for each deployed agent.
Related insights
