What HR Leaders are Actually Thinking About in 2026

Across sectors, the same admission kept surfacing: the hard part is not longer—or was never really—the AI. A Chief People Officer at a national dental services company put a number on it: 60-70% of an AI rollout was about change management and process engineering, not the technology itself.

The remaining questions — job redesign, and where the ROI actually shows up — follow from that directly: lighter reporting loads, faster charting and notes, and organizational design that reflects the new division of labor.

A People and Culture leader at a mid-size benefits provider, with around 1,200 employees, described a five-year transformation project run with an outside partner and staffed by a dedicated transformation department pulling in subject matter experts from across the business — notably led by someone from the business side, not IT. That structural choice came up more than once as the difference between a rollout that sticks and one that stalls.

You have to start from business outcomes first, and test individual use cases against those outcomes rather than deploying tools and hoping value follows, according to a people leader at a large construction company with roughly 6,000 employees. This keeps the AI conversation anchored to what the business needs, not just what the technology can theoretically do.

Not every organization in the room was choosing its own pace. An HR and security leader at a federal government agency described a sector defined by risk aversion, now under real pressure to accelerate alongside a significant workforce reduction happening simultaneously. She named the change being perceived bluntly, like “taking two steps forward and three steps back," with real fear and real pushback from employees living through both the transformation and the headcount reduction at once.

That fear showed up again, in a different register, at a national newspaper. The newsroom's anxiety has two distinct sources: concern that AI tools might be trained on sensitive or confidential material, and a wave of downsizing that's made any AI conversation feel adjacent to job security. Journalists are concerned about the integrity of their craft on various fronts, from fact-checking through to reporting a story objectively.

But some journalists are also genuinely curious what AI could do for their research, trust and data questions aside. The clearest point from that HR leader wasn't about the technology at all:

"It's on managers to create the space for people to explore and learn AI — and to drive the adoption and guide their employees through it."

A theme that surfaced across multiple sectors, and rarely makes it into public discussion of AI adoption: cost control at the token level.

A People and Culture leader at a fuel logistics company described a standing cross-functional committee — HR, Finance, and IT together — built specifically to manage this. The company has standardized on two major AI assistants, primarily for personal employee productivity, but adoption still carries real hesitation: employees are wary of the tools affecting their own job security, even as the organization works to reframe AI as something that helps them do their jobs better. Underneath that hesitation sits a practical worry — token costs and AI spend sprawling faster than anyone budgeted for.

A senior HR and ethics leader at a large energy utility described the same pressure from a different angle: understanding who the organization's AI “superusers” are, and the velocity at which they're adopting new tools, matters as much as the tools themselves. Token cost is a significant pain point and a real source of internal friction — one more reason, she noted, that supporting workforce redesign now requires mapping skills before and after the shift, not just deploying the technology and waiting to see what happens.

Another HR leader raised a related concern from the data side: AI “superusers” building their own connectors and finding their way to sensitive employee information faster than governance can track. Her answer wasn't a policy document — it was proactive engagement with IT, built before the worst case happens, not after. A People leader at a healthcare organization framed the same problem from the culture side: the organizational skills — culture, skills, and governance together — needed to scale AI-driven innovation responsibly.

An HR leader at a senior-living company took a more foundational approach: align stakeholders early, particularly those who need to commit real time to any rollout, before scaling anything. The company has already implemented agents in accounts payable and receivable, along with an HR chatbot for knowledge-base questions — deliberately starting with contained, well-bounded workflows rather than an enterprise-wide push.

Several leaders in the room described a version of the same shift: HR's role is moving from executing workflows to designing the architecture those workflows sit inside.

A Chief People Officer at a global tech company offered the sharpest version of this: "Everybody thinks everybody should be doing everything with AI, but we've realized we don't need all 3,000 people using AI for everything. We need to figure out the right workflows and structures and use cases to do so." Her argument: HR has a role no other function can substitute for — discernment, deciding what to scale and what not to scale yet.

A Chief People Officer at an agriculture and produce company named a practical constraint behind that shift: agentic AI engineers — the people who actually need to build and manage these systems — are hard to hire, and the cross-functional questions involved (who owns the business case, where R&D fits) don't have settled answers yet.

A People leader at a steel manufacturer described the emotional layer underneath the technical one: task automation shows clear value and productivity gains, but fear shows up at both the employee and manager level, and a real part of the job now is actively working to decrease that fear rather than assuming it resolves on its own.

An HR executive at a large energy utility raised a scheduling problem as much as a strategic one: the organization is in the middle of a major ERP implementation, and figuring out how to layer an AI strategy on top of an already-massive transformation is its own distinct challenge, separate from the AI question itself.

A CHRO at a major bank reframed the hiring question entirely: "who have we not hired?" Every systems integrator partner is already engaged, with a deliberate focus on federated oversight and governance — spreading accountability across the partnership network rather than centralizing it in one team.

And at a payroll and HR services company operating across many countries, one leader described tailoring the approach by geography — what fits varies from market to market — while rolling out an agentic AI tool specifically for performance management, a workflow most organizations haven't yet touched with agentic tools at all.

One HR leader, building an AI enablement center to address the scaling challenge nearly everyone in the room named in some form, offered the line that best captures where this is heading: in headcount planning discussions, leaders are already being asked, "if you have 8 humans and 2 agents, that's your team of 10 — why do you need more humans?" No one in the room had a fully drawn AI-plus-human org chart yet. But the question itself shows the org chart is already being redrawn in practice, one planning conversation at a time.

The consistent advice from leaders further along: start with leadership adoption, not employee adoption. The mindset shift has to happen at the top first — from "AI will take my job" to "AI is an enabler that will make me more efficient and let me take on new skills" — before it can spread credibly through the rest of the organization. Transparency was named repeatedly as the other non-negotiable: being honest that some jobs will genuinely go away, while new jobs and new skills get created in their place, rather than promising that nothing will change.

An HR leader at an insurance brokerage connected this directly to career pathing and change management — treating the human side of the transition as a designed program, not a byproduct of the technology rollout. And an HR function at a major bank, running roughly 1,800 people, has already done substantial work mapping which skills and roles are most exposed to AI, then building upskilling programs starting deliberately with leaders first.

The line that closed the loop for us came from one leader responding to the room's broader questions: "Technology is only going to become less expensive and more fully available. The only competitive advantage that remains is our people." Every organization in that room, regardless of sector or size, was working some version of that sentence into their actual operating model.

What's notable about this room wasn't the diversity of sectors — it was how similar the underlying questions were, from a 1,200-person benefits provider to an 1,800-person bank HR function. The technology decisions varied. The organizational ones didn't: where governance has to move as fast as adoption, where token costs become a real budget line, and where the hardest work is redesigning jobs and building trust, not deploying software.

That's the work Ema exists to be part of. If your team is working through workflow discernment, governance built for how AI actually gets used, or what redeployment and upskilling look like at scale, we'd welcome comparing notes — or exploring how organizations like Wipro have approached the same redesign question at a much larger scale.