Most organizations I work with are still treating AI in HR as a productivity story. Faster recruiting, faster onboarding, faster training. That story is real, but it is half the work. The other half is structural, and most HR functions are not yet set up to lead it.
I engaged with three companies last year that asked me to help with their AI strategy and transformation. Same year, same questions, very different starting points.
In the first, HR and L&D were leading the AI conversation. They had read the field, took an active part in the AI task force leadership had set up, and brought me in to help shape the program. The HR team sat at every working session and the program had a concrete workforce-impact line on it.
In the second, engineering led the way in building the company's AI capability. They didn't wait for anybody to tell them to do so; it grew from engineers seeing their work change. When the head of L&D asked to be part of the strategy work, the engineering director was polite: "We've got this. Don't disturb the team. We'll loop you in when we need a training plan."
In the third, the AI strategy was already written when I arrived. The HR team was not part of it, and key HR aspects were missing from the roadmap or allocated budget. Nobody had thought to invite HR. It had genuinely never occurred to them. HR was a service function. This was an AI initiative. The two would meet later.
These three positions, leading, pushed-aside, and not-invited, are not curiosities. They are the three states most organizations are in when AI hits the HR function. They predict, with embarrassing accuracy, what the next two years of the transformation will look like for each one. What separates the first from the other two is a structural choice: did the company set HR up to lead two different streams of work at once?
I write more about why HR has to be ready to take that seat in this issue's leadership post on Earth's Best Employer in the age of AI. Here I want to walk through what the work actually is once HR is in the room.
Mandate 1: Transform HR itself
This is the one most leaders are already comfortable with. Recruiters use an AI-powered system to surface candidates a resume search would have missed. L&D builds personalized learning paths against a real skills graph rather than a generic curriculum. Performance management stops being an annual ritual and becomes continuous coaching. Compensation gets benchmarked against external labor market data in real time. Managers stop guessing about who's at risk of leaving and start getting actionable retention signals.
Mandate 1 is the productivity-and-value side of the work. It is real, it is necessary, and it has measurable cycle-time outcomes. The Findem case study at RecruitMilitary in this issue's tools section is a clean example of what good Mandate 1 work looks like with real numbers behind it.
Mandate 2: Redesign the organization
Mandate 2 is harder, more strategic, and spans the whole organization rather than HR alone. The HR function exists to optimize the workforce, including the AI agents that are now joining it. Ask 100 Chief HR Officers, and I doubt a handful will see it this way yet.
What does Mandate 2 look like in practice?
It looks like HR helping decide what a job is in the first place. Josh Bersin's bet is that vertical job families, sales and marketing and engineering and support, are an industrial-era artifact, and that companies will increasingly operate as horizontal pods that form around customer outcomes and dissolve when the work is done. The customer doesn't see the verticals. The customer sees the horizontals. Someone has to design the human side of those pods. That someone is HR.
It looks like HR setting the rules that govern what the agentic HR system is and isn't allowed to do. The talent-redeployment super-agent that can model how to take ten percent out of a business unit's labor cost is going to exist. Whether it recommends terminations, redeployments, pay freezes, or some combination depends entirely on what HR has told it to value. If HR isn't shaping those rules, somebody else is, and that somebody else is probably an engineer.
And it looks like HR owning the regulatory perimeter. The EU AI Act treats most HR-related AI as high-risk. NYC Local Law 144 requires bias audits. GDPR governs employee data. Works councils have a vote in much of Europe. Someone in the room has to know it cold and have the authority to act on it. If HR doesn't own that perimeter, the lawyers will, and lawyers default to no.
Two mandates, two rhythms
The two mandates run on different rhythms. Mandate 1 is operational. It lives inside HR, has measurable cycle-time outcomes, and touches existing processes one at a time. Mandate 2 is strategic. It lives across the org, shapes future operating models, and requires HR to sit alongside the CTO, the COO, the CISO, and General Counsel.
Most HR functions try to do both with the same team and the same calendar. That is how Mandate 2 dies. The operational work always wins on a busy week, because the operational work has a deadline. Mandate 2 has to be staffed and scheduled deliberately, even if it is the same HR leaders sitting in different rooms.
The OPSWAT playbook: change management is HR's craft
Jennifer Ert, Chief People Officer at OPSWAT, ran into both mandates directly and is worth listening to. OPSWAT is a $200M cybersecurity company protecting critical infrastructure: 99% of nuclear plants and a long list of water-treatment facilities. 1,200 employees globally, 500-plus engineers, distributed across Vietnam, Hungary, Romania, and the Americas. When her CEO asked her to lead the company's AI transformation, starting with engineering, her own HR leadership team's first reaction was that it was "too technical." Then they realized something Ert names cleanly in her recent appearance on the Future Proof HR podcast: "It's a change management effort. And it's all the things we knew how to do."
That observation, change management is HR's craft, is the most underused argument for HR's seat at the AI table. John Kotter's eight-step change framework, taught in every HR L&D program for decades, maps almost cleanly onto what an enterprise AI rollout actually demands. Establish urgency. Form a guiding coalition. Build a vision. Communicate it. Remove barriers. Generate short-term wins. Sustain acceleration. Anchor the change. HR has been doing this for decades. The AI transformation is just a high-stakes special case.
Ert's playbook for setting up the work is worth copying. She brought her global HR leadership team together in person. Trust and alignment compress when people are in the same room. She invited the CTO, COO, and CISO to present to that HR team, not to brief them but to surface what the business actually needed. From that meeting they built a project plan with four streams: talent acquisition, organizational design, culture, and embedding the change in OPSWAT's values and leadership principles. Engineering became the first internal customer because that is where the company's product impact compounds. The AI Academy they built for engineers is now rolling out to every other function.
Read those four streams again. Talent acquisition. Org design. Culture. Values and leadership principles. Those are existing HR competencies, applied to a transformation that needs them all at once. The team to run Mandate 1 is largely the team you already have, freshly upskilled on the AI tools they will use and audit. The team to run Mandate 2 is the same HR leaders in a different posture: at the executive table, paired with engineering and product leaders, working on the design of the org rather than its operations. Same people, different rooms, both rooms required.
A diagnostic for your own team
Ask the leader of HR, and yourself, three questions:
- Who is leading the work to transform HR itself?
- Who is leading the work to redesign the organization around hybrid human-AI teams?
- What is the change-management plan that connects the two?
Two different mandates, two different team postures, both starting now. If those questions all land on the same overworked CHRO, you don't have enough HR to lead the change.
Your action step
In your next leadership team meeting, take 30 minutes and put the three diagnostic questions on a whiteboard. Name a leader against each. If the answers are vague, your AI strategy has a workforce blind spot, and that blind spot is where the transformation will fail. Pair the diagnostic with a concrete read of the Findem / RecruitMilitary case and a working definition of skills-based organization so the leadership team has a shared vocabulary.
If you'd like help designing the dual-mandate operating model for your HR function, or want me to run a working session with your leadership team on this, I'd love to help.
Sources
- Jennifer Ert, OPSWAT, Future Proof HR podcast
- Josh Bersin, HR 2030 vision (pod structures, agentic HR systems)
- John Kotter, Leading Change (eight-step framework)
- EU AI Act, NYC Local Law 144, GDPR (HR-AI regulatory perimeter)