Quick word before I get into it. This isn’t AI written slop, and it isn’t a think piece I’ve cobbled together from whatever’s trending this week. It’s genuinely my own view, built from a lot of conversations with HR and business leaders over the past two to three years, people who are right in the thick of trying to figure this out. I don’t have the golden bullet. Anyone who tells you they do is selling something. But I do have a view, and I think it’s worth sharing.
So here it is…
There’s a conversation happening in pretty much every HR leadership team right now, and it usually goes something like this. “Where can we use AI? What can we automate? How much can we save?”
Fair questions. But I think they’re the wrong place to start, and starting there quietly nudges everyone towards the least interesting, least valuable version of what’s actually possible.
Here’s what I keep coming back to. If AI really does take on a decent chunk of the transactional and advisory work that HR has been built around (the query handling, the first line advice, the admin, the reporting), then the interesting question isn’t how much you can save. It’s what HR gets to become once it’s no longer buried in the work that’s defined it for the last thirty years.
And honestly? I think the answer should have HR leaders rubbing their hands together, not lying awake at night.
Let me explain what I mean about HR being built around the wrong thing.
Look at how most HR functions are actually put together and you’ll see they’re organised around process. Recruitment over here. Reward over there. Employee relations, learning, HR ops, and an HRBP layer stretched across the business doing its best to hold the whole thing together. That didn’t happen by accident. It exists because the work was largely transactional, and transactional work needs people organised around processes to keep it moving.
So the shape of HR is, when you really look at it, a bit of a monument to the kind of work it used to do.
Now picture a real slice of that transactional load shifting over to AI. Not all of it, not overnight, but enough that the whole organising logic starts to wobble. Suddenly the structure that made total sense is built around work that something else is increasingly doing.
This is the moment I think most organisations are going to fluff. They’ll take the efficiency, quietly trim a few roles, bolt the AI onto the existing shape and call it transformation. They’ll treat the whole thing as a cost story. And in doing that they’ll waste the biggest opportunity HR has had in a generation. Which would be a real shame.
Because here’s what taking AI seriously actually asks of you. It asks for a stronger strategic capability, not a smaller function.
When the transactional layer shrinks, what’s left (and what suddenly becomes possible) is the genuinely hard, genuinely valuable stuff. Proper workforce planning, the kind that looks years out rather than just plugging next quarter’s gaps. Organisation design that actually moves the needle on how the business performs. Talent strategy that’s joined up with where the company is heading, not just who handed their notice in last week. OD. Skills and capability architecture. The human, slightly messy work of leading people through change that, let’s be honest, isn’t going to slow down any time soon.
This is the work HR has been saying it wants to do for years and rarely had the room to do properly, because it kept getting dragged back into the churn.
AI is the thing that could finally change that. But only if the freed up capacity gets reinvested into higher value strategic roles rather than just stripped out of the cost base. That’s the real choice sitting in front of every HR leader right now, whether they’ve named it that way or not. Do you use this moment to make HR smaller, or to make it genuinely better?
The organisations that get this right are going to look pretty different in a few years. Fewer people running process, a lot more people doing work that needs judgment, relationships, design thinking and a bit of strategic backbone. A function organised around the value it creates rather than its admin domains. The HRBP role finally given some real teeth instead of being stretched three ways at once. And new roles that barely exist yet, the workforce strategists and capability architects and the people who can actually design how humans and AI work side by side.
That’s not a smaller HR. It’s a braver one.
And the slightly awkward truth is that getting there is an organisation design problem far more than a technology one. The tech will turn up. It always does. The harder work, the work that actually decides whether any of this lands, is rethinking the shape of the function, the roles inside it, the capabilities you need to build or buy, and the route from where you are now to where you want to be. Funnily enough, none of that is in the vendor’s slide deck.
So if you’re in an HR leadership team having the “where can we use AI” conversation, I’d gently nudge you towards a different starting point. Not “what can we automate,” but “what do we actually want this function to be on the other side of all this, and what would we need to redesign to get there.”
Harder question. Also the one that matters.
That’s the work I find genuinely interesting, and it’s where I spend most of my time, sitting in the space between AI, operating model and the human reality of making it all actually work. Like I said, I haven’t got the golden bullet. But I’ve got a view, and I’m always up for a good conversation if you’re wrestling with any of this.
More soon.