Most HR org structures are still built for a 2005 world, while work is changing like it's 2026.
Josh Bersin floated the headline: Core HR headcount could fall by ~30%.
I don't think the number is the point. I think the operating model is the point.
That 30% happens if we change nothing and just bolt agents onto yesterday's machine.
Shared services doing transactions. COEs doing programs. HRBPs trying to be strategic in the cracks. A transformation team off to the side.
Sound familiar?
We'll automate the transaction layer and call it productivity. Translation: cuts. And it's already happening.
The takeaway from Gartner
I sat in a Gartner webinar recently — To Lead a Successful HR Transformation, A Good Process is Not Enough.
The takeaway wasn't get a better process. It was: process isn't the differentiator. Ownership and work redesign is.
And Gartner's broader AI message is blunt:
- Most AI still doesn't deliver measurable value (1 in 5 hits measurable ROI; 1 in 50 is disruptive)
- If you're not redesigning work end-to-end, you're basically buying expensive demos
What I'm pressure-testing
HR shouldn't be an operating model. It should be an operating system that is completely redesigned.
Two buckets. One backbone.
1) People Platforms (Shared Services, redefined)
Payroll, benefits, total rewards, wellbeing, recruiting, people analytics and workforce intelligence, internal mobility — but run like platforms: with product ownership, data, AI built into workflows, and customized exception handling.
2) Work Design + Adoption (Transformation Services, but for real)
Change management and comms, learning in the flow (not courseware), manager enablement, performance and feedback, EX design, HR technology — but measured on friction removed and work redesigned, not programs launched or number of attendees at a workshop.
The backbone: Transformation Management Office + AI governance
Nobody wants 100 agents from 100 vendors and a pile of risk no one owns.
This is the difference between HR as cost and overhead — and HR as a growth catalyst plus risk control system.
Pressure test it with me
I'm pressure-testing this operating system and want feedback.
What's missing? If you had to kill one part of today's HR structure first… what goes?
Building or rebuilding your HR operating model? Let's talk.