The workforce research pouring out of the major firms in 2026 converges on one uncomfortable idea: A job title may still name one person, but the value behind it is increasingly produced by a network of humans and AI agents whose contributions are hard to separate. That reframing sounds abstract until it reaches the hiring process, where every instrument companies use to evaluate talent—the resume, the years of experience, and the list of past titles—was built to measure individuals working alone. Compounding matters is the fact that agentic AI is still waiting on its proverbial iPhone moment, the point at which the technology becomes intuitive enough that anyone can pick it up and use it out of the box. If value now comes from how well a person orchestrates machines, employers are trying to staff a hybrid workforce with tools that can't see the thing that matters most.
McKinsey puts the point directly, asserting that "the true unit of value is no longer a single role, but a coordinated system of humans and agents delivering outcomes together," and calling for performance management to shift from individual output to system effectiveness. BCG's recent survey assigns numbers to how fast this is arriving, finding that 30 percent of organizations have integrated AI agents into workflows, more than double the prior year, even as it warned that operating models haven't caught up.
The workforce is being restructured faster than the tools to measure it
The people who study this shift for a living describe a restructuring already underway. "Agents are coming much faster than most people expect," Rishad Tobaccowala, the former chief strategist at Publicis, told the audience at a New York future-of-work conference. He sketched a workforce in which the traditional full-time employee splits into "fractionalized employees, AI agents, or part of a marketplace." When the composition of a team changes that fundamentally, the question of how to evaluate the humans in it changes with it.
Industry analyst Josh Bersin frames the same transition as a reinvention of the function that does the hiring. Agentic applications, he writes, are creating "a new and vital role, building, stitching together, and architecting the AI agents that automate" work, a capability that looks nothing like the credentials on a conventional resume. He's blunt about the stakes of getting the human oversight wrong. "In an automated system operating at scale, a single error in a recruiting algorithm could affect thousands of employees instantly," he noted.
It's too early to name the skills, because the interface hasn't arrived
For companies trying to hire against that moving target, some practitioners caution that the technology hasn't stabilized enough to define the competency yet. "Agentic AI has not yet had its iPhone moment," said Wael Sabra, founder and CEO of the global talent platform SiiRA. Until that arrives, he believes, prescribing the skills a worker needs to collaborate with agents is premature. "I really believe it's too early to tell exactly what skills humans need to have to interact with AI agents," Sabra said adding that the volume of talk outpaces the reality. "There are a thousand times more people talking about creating agents than there are actually creating agents."
That gap makes screening for agent proficiency treacherous, because the proficiency barely exists in practice. A candidate can claim fluency with a technology that most organizations are themselves only piloting, and no resume convention exists to separate genuine capability from familiarity with the hype.
Maturity is the thing to assess
Sabra moves the question away from the candidate and toward the organization. He describes an AI maturity model in which the native stage is simply that individuals use AI. "If your maturity framework is 'all our developers use AI,' that's level one," he said. "This is where expertise-based, skill-based is totally fine, because you still need the operator behind the screen knowing what they're doing." The harder stage is when AI is embedded across a workflow, and there the evaluation problem sharpens.
His example is specific: a designer who can build in a tool but can't connect an AI assistant into it isn't useful to a company operating at the embedded level. "If they have experience building designs but don't know how Claude could help you within Figma, that type of designer is not going to help the organization," Sabra said. The relevant capability is not a line on a CV but a fit with how mature the organization's own workflows have become, which is precisely what McKinsey means by clarifying where humans, agents, or hybrids work best.
Governance is the gap the resume hides
Underneath the evaluation problem sits a governance one that the research suggests most companies have barely confronted. BCG found that half of the organizations it surveyed lack clear governance for managing mixed human-agent teams, and only 36 percent of employees feel adequately upskilled for the shift. Bersin's warning about a single algorithmic error scaling instantly is the same risk viewed from the top: when a system rather than a person produces the work, accountability for the reasoning inside it becomes the thing leaders have to design for, and the thing a resume will never surface.
For technical leaders, the theme is consistent. Evaluating talent for a hybrid workforce is less about finding people who list agent experience and more about finding those who can design, supervise, and stand behind these systems as they scale. That capability doesn't fit the resume, which is why the consultancies, analysts, and platforms arrive at the same conclusion: the organizations that come out ahead will redesign how they assess talent as fast as they redesign how the work gets done.
