Leading the AI transformation in People

Almost every AI failure inside a People function is, on closer inspection, a change failure. The tools work. The workflows are sound. The capability could be built. What broke is the human side — fear that was never named, sequencing that asked too much too soon, leaders who confused announcing change with leading it.

This is the part of the work most CPOs find hardest, because it is not about AI at all. It is about leading a meaningful transformation while the rest of the business is also changing, with a team that is rightly anxious about what this means for them.

The good news is that there is a sequence that works. It is not novel. It is the same sequence that works for any serious change. It just has to be done — properly, and in order.

Start by naming the fear

The single most common opening mistake is treating AI as a productivity story. "This will free us up to do more strategic work." People hear that sentence and the next thought is: or to do the same work with fewer of us. They are not wrong to think it.

The leaders who get the next twelve months right open the conversation differently. They name the fear before anyone else has to.

A version that works:

I want to be straight with you. AI is going to change what this team does. Some of the work we do now will be done by systems within a year or two. Some roles will look meaningfully different. I do not have a complete map. What I can tell you is that we are going to build this together, the team is going to grow capability faster than the work shrinks, and nobody on this team is going to be surprised by a change to their role. We will talk about it openly as we go.

That paragraph, said with the lights on, does more for adoption than any tool rollout. The team can now have the real conversation. They can stop working out, in private, what management is "really" planning.

The fear is also yours. CPOs who pretend they have it all figured out lose credibility within a quarter. The teams that end up most engaged are the ones whose leaders said, early: I am learning this in public, the same as you.

The sequence that works

Five phases. None can be skipped without paying for it later.

Phase 1: Personal capability (months 1–2). The CPO and the leadership team build their own working knowledge. Not a one-day course. Real reps. They use AI in their actual work — board prep, calibration, comms — for two months before asking anyone else to. This is non-negotiable. A leader who cannot describe what they have personally built with AI has no standing to lead this change.

Phase 2: A small visible win (months 2–4). Pick one workflow. Bounded. Painful. Visible. Build it. Ship it. Tell the story of what changed and why. Not a pilot deck. An actual thing the team is using on a Wednesday. The point is not the workflow. The point is to demonstrate, concretely, that this works inside our team, with our tools, on our problems.

Phase 3: Champions and shared workspace (months 3–6). Three or four champions named. The shared workspace set up properly. The first batch of people start using it for daily work. Capability begins to compound.

Phase 4: Function-wide rollout (months 6–12). The patterns the champions have built get rolled out across the function. Onboarding for new joiners includes the workspace from day one. The team's internal language changes — people stop asking "should we use AI for this?" and start asking "which of our patterns fits this?"

Phase 5: Redesign (months 12+). The conversation about team shape becomes real. Roles evolve. Some shrink, some appear. The work of the function looks measurably different from where it started.

The phases overlap. They are not crisp gates. But the order matters. Skipping Phase 1 produces the most expensive failures. Skipping Phase 2 produces strategy decks that nobody believes. Skipping Phase 3 produces hero capability that does not survive the first departure.

The four resistance patterns and what to do about them

Resistance is not a problem to be eliminated. It is information. Four patterns recur, and each has a response that works.

The Sceptic. "I tried it once, it was wrong, this is overhyped." Usually a senior, experienced operator who is not wrong about much. The response is not a deck. It is a one-on-one demonstration of a real workflow that solves a problem they actually have. Sceptics convert when they see leverage in their own hands; they never convert from a presentation.

The Worrier. "What about bias? What about confidentiality? What about quality?" These are the right questions. The response is to take the worries seriously and answer them with substance — show the governance work, the data classification, the human-in-the-loop checkpoints. Worriers, treated well, become some of the strongest advocates because they have stress-tested the work.

The Performer. "I am already using AI loads, I am ahead of everyone." Often actually true. The risk with performers is that they become a single point of capability; the work does not transfer. The response is to channel them into champion roles where their job is to spread the practice, not to hoard it.

The Quiet Quitter. "Sure, sounds great." Says yes in meetings, does nothing between them. The response is operational, not motivational: small, specific, time-boxed asks tied to their actual work, with a follow-up date. Either the work happens and the engagement follows, or the pattern surfaces clearly and you can have a different conversation.

The mistake is treating all four with the same intervention. They need different things.

What stops the transformation working

A short list, in rough order of frequency.

No protected time. Champions and builders are expected to do the new work on top of their existing work. Within six weeks, the new work disappears. Twenty per cent of the week, written down, defended by the CPO, or it does not happen.

Tool-shopping as a substitute for building. Months spent evaluating platforms is months not spent building things with the platforms you already have. Pick something good enough, start, switch later if you have to. The cost of the wrong tool is small. The cost of a year of evaluation is huge.

Communication that outpaces reality. Announcing an "AI-first People function" before anyone has shipped anything. The team learns that the words and the work are not connected, and stops trusting either. Build first, talk after.

Leaders who delegate the change. "I have asked Sarah to lead our AI work." Sarah, however good, does not have the authority to redesign roles, protect time across the function, or hold the line when budgets get cut. The CPO has to lead this personally, with Sarah as a partner, not a proxy.

Treating it as a project with an end date. AI in the People function is not a programme that finishes in Q3. It is the new shape of the work. Programmes that are framed as having an end date stop getting attention the moment the next thing arrives.

What to measure

Resist the urge to measure usage. Number of prompts, number of users, hours of training delivered — these are vanity numbers. They tell you who is busy, not who is getting leverage.

Three things are worth measuring, all qualitative-leaning.

Workflow count and depth. How many workflows does the team actually run on AI infrastructure today, end to end? How many of them survived their original builder leaving the project? This is the closest thing to a real adoption number.

Time reallocation. Has the percentage of People-function time spent on coordination and drafting actually fallen? Has the percentage spent on judgment and partnering actually risen? If the time profile of the team has not moved after twelve months, the transformation has not happened, regardless of how many tools you have bought.

Confidence and capability. Ask the team, twice a year: can you confidently use AI in your daily work? Could you teach a new joiner how this team uses AI? The shift from "no, but" to "yes, and" is the real signal.

The board wants ROI; show them the value piece for that. The team's progress lives in the three numbers above.

The hardest part

The hardest part of leading this transformation is staying with it when the headlines move on. AI is currently the most-discussed topic in any leadership conversation; in eighteen months it will be background, replaced by whatever is next. The leaders who keep building through that quiet period are the ones whose teams end up genuinely transformed. The ones who treated it as a campaign find their work undone within a year.

This is the same lesson as every other deep change in operating practice. The headlines fade. The grain remains. Lead with that in mind from the start, and the transformation has a chance of becoming the new shape of the team — not a chapter in a strategy deck nobody opens any more.

What this connects to

Auto-recommended next reads in the People Ops cluster, ranked by shared concepts and headings:

• An AI enablement operating model for People leaders
• Measuring AI value in People Ops
• A workflow assessment framework for People Ops
• Designing values that stick