Something important changed in 2025; many organizations started talking about AI agents as part of the workforce. McKinsey described “the digital workforce” as a near-term reality in which humans work side by side with AI agents, while Salesforce explicitly framed this shift in terms of “the human + agent workforce”.

Once agents are discussed as part of workforce design rather than as tools, the question becomes which forms of digital labor belong where, under whose supervision, and with what governance.

In January 2026, Axios reported that Mark Zuckerberg described Meta’s push toward “AI-native tooling” and said the company was beginning to see projects that once required large teams being completed by “a single, very talented person”. The point is that leading companies are now explicitly redesigning work around the leverage AI creates at the individual and team level.

Salesforce’s own research made the same shift visible from the HR side. In its May 2025 announcement for Agentforce HR Service, Salesforce said “eight in ten HR chiefs say the workforce will be made up of humans and agents in the next five years”.

When a company buys software, the decision criteria are usually functional: capability, integration, security, cost, vendor stability. That process often sits with IT, procurement, and architecture teams. When a company appoints a senior executive, the criteria are materially different: judgment, fit, leadership range, context, risk, and the ability to perform inside a specific organization. That process sits with the CHRO, CEO and the board.

AI agents are still bought and built like software. But increasingly they are being expected to perform like labor.

McKinsey has argued that strategic workforce planning is moving away from static role-based thinking toward “dynamic, activity-based models that specify which tasks are performed by humans, AI agents, or hybrid teams”. It has also argued that managers will increasingly “supervise, evaluate, and develop agents”, not just people. Those are a redesign of workforce architecture itself.

The default response in most organizations today is still build or buy. Either build agents internally around proprietary processes and data, or buy an enterprise platform and configure it.

Both build and buy approaches are heavily shaped by technical logic. They prioritize what the system can do, how it fits the stack, how quickly it can be deployed, and how securely it can be governed. Those questions matter. But what is the right mix of human and digital labor for this organization, in this function, at this level of risk, with this leadership culture?

That is not purely an engineering question nor a procurement question either.

We have seen this before in executive hiring. Organizations used to over-index on functional fit: sector knowledge, title match, operating experience, geography. Over time they learned that those factors alone do not predict success. The quality of the match depends on context: leadership model, strategic situation, team composition, board expectations, decision velocity, cultural tolerance for ambiguity. In other words, selection got more sophisticated when organizations realized the profile could not be evaluated in isolation.

The same will happen with agents.

An agent that is technically strong may still be a poor fit for a given operating context. An agent that performs well in one environment may create unacceptable risk in another. A platform that looks compelling in demo form may fail once it interacts with real workflows, real accountability structures, and real governance constraints. ADP’s March 2026 launch of AI agents within ADP Marketplace is one sign that the market is already moving beyond a raw-tool mindset toward more curated, trusted ecosystems. That does not solve the problem, but it does confirm that selection and curation are becoming part of the value layer.

This is why a new leadership discipline is emerging: agent portfolio design.

The analogy is closer to portfolio construction than to conventional software selection. A strong portfolio is a composition built for a specific objective, with explicit trade-offs around return, concentration, resilience, and risk. The same logic applies to human-agent workforce design. The question is whether the overall mix of people, agents, workflows, controls, and decision rights is appropriate for what the organization is trying to achieve. And that requires a different kind of leader.

McKinsey’s recent work makes clear that HR is being pulled closer to the center of this shift, because work itself is being redefined. Its framing of hybrid, co-intelligent teams implies that workforce decisions can no longer stop at human headcount planning. They now extend into the design of work across humans and agents.

The market is starting to name this change more directly. In December 2025, Metaview predicted the rise of “agent managers” as a management role, reflecting that some people will be materially better than others at directing and extracting value from fleets of agents. Whether or not that exact title sticks, the underlying point is sound: new management capability is being created.

The organizations that handle this well will not be the ones that merely adopt the most agents. They will be the ones that make better decisions about where agents belong, where they do not, how they should be governed, and which leaders are capable of designing systems in which human and digital labor reinforce one another rather than create noise, fragility, or hidden risk.

That has direct implications for leadership hiring.

The strongest future CHROs, functional leaders, and enterprise transformation leaders will be those who can think clearly about work decomposition, decision rights, managerial oversight, risk boundaries, escalation paths, and the practical conditions under which hybrid teams outperform purely human ones. McKinsey’s work on the agentic organization consistently points toward orchestration, redesign, and continuous adaptation as the real leadership burden.

In other words, the emerging question is not only who can build agents. It is who can build a workforce that includes them.

That is a different search problem.

In executive search, the hardest part has been definition: knowing what the role really requires before the market has a settled template for it. The same applies here. Most organizations do not yet have a mature specification for the leaders who will shape their human-agent workforce. The title may still be unclear, but the need is already visible.

The companies that treat this as a narrow tooling decision will get narrow tooling outcomes. The companies that treat it as a workforce architecture question will be in a stronger position to build something more durable: a model in which people, agents, and management systems are designed together rather than bolted together after the fact. That is where the next real leadership advantage is likely to come from.

Sources