Tess Hollis: Felix, hey — quick thing before we get going: if someone told you a single tweet with six likes was the founding document of a major industry debate, what would your prior be on the underlying data?
Felix Ortiz: I mean — honestly, pretty low? Depends who posted it I guess.
Tess Hollis: It's Matt Schulman. CEO of Pave — the compensation analytics company. October 13th, 2026. He claims Pave's own data shows HR headcount fell from 5.3% to 4.3% of the pre-AI tech workforce. No named analyst has validated it — not Josh Bersin, not SHRM, not Gartner. And when Perplexity, Claude, and Gemini all independently searched for the 5.3%-to-4.3% figure in any Pave report or publication? Nothing.
Felix Ortiz: Okay — so the number might be real and just unpublished, or it might be... something else entirely.
Tess Hollis: That's the split we have to sit in. Because the question isn't really 'is AI hurting HR' — it's whether we even have evidence to start that conversation, or whether we're building on a foundation that's just a post.
Felix Ortiz: Six likes. Two replies. That's a remarkably quiet ground zero for something this loud.
Tess Hollis: And the quiet is the tell — but here's what I keep snagging on. Even if the number is real, the ratio can't prove what Schulman says it proves.
Felix Ortiz: Okay — what do you mean it can't prove it?
Tess Hollis: Picture ten people in a room. Two are HR. That's 20%. Then eight engineers walk in. Now you've got eighteen people, still two HR — that's 11%. HR didn't shrink. The room got bigger.
Felix Ortiz: Oh. Yeah — wait, that's exactly what Pave's own data shows, right? AI engineering's share of tech employees went from 0.010% to 0.185% between Q4 2023 and Q2 2026. That's like — I mean, that's roughly 18x in under three years. The room isn't just bigger, it's a different room.
Tess Hollis: So the denominator is basically exploding. Which means Schulman could be looking at a ratio drop — 5.3 to 4.3 — and calling it an HR crisis when it might just be... math.
Felix Ortiz: Right — and we genuinely don't know which story it is. Nobody said 'we cut 200 HR people.' We're looking at a ratio and, yeah, working backward from that is — I mean, that's doing a lot of interpretive work that the number doesn't actually license.
Tess Hollis: And it gets stranger when you look at who specifically is getting squeezed — that part, the Washington Post data on age and role level, actually breaks this open in a direction Schulman didn't go.
Felix Ortiz: The Washington Post piece — October 13th, same day as Schulman's post — actually names it. AI-exposed occupations shrinking specifically for workers aged 22 to 25. Not experienced professionals. Junior people. The headcount isn't disappearing from HR broadly, it's disappearing from the bottom of the org chart.
Tess Hollis: And that bottom layer is exactly what TriNet just automated.
Felix Ortiz: Wait — TriNet announced that on October 13th too?
Tess Hollis: Same day. New AI-powered platform capabilities for SMBs — embedded recruiting, leave management, onboarding tools. The exact job description of a 23-year-old HR coordinator hired in late 2025 to process paperwork and schedule interviews. That's not a coincidence, it's just... the timeline collapsing on itself.
Felix Ortiz: So it's not that AI is being automated out of HR — it's being automated into HR. And the person who gets displaced is the one who was doing the tasks the software just absorbed. Which is — I mean, Josh Bersin's been saying something like this, right? That headcount shrinks but the function actually shifts toward higher-value work?
Tess Hollis: Bersin's argument is specifically that AI reduces headcount while moving remaining professionals toward strategic roles. Caroline Hewings at Gartner pushes it further — organizations have to redesign work in real time, not just update job descriptions. Which means the function isn't static while this happens.
Felix Ortiz: And WorkTech Desk's number — 210,000 workers laid off in events where employers cited AI by mid-September 2026 — that's real, but it doesn't isolate HR at all. We can't actually pull HR out of that figure.
Tess Hollis: Which is the whole problem. The stratification story — junior out, strategic up — fits every data point we can actually verify. But Schulman named it as pressure across the board. Those are different diagnoses entirely.
Felix Ortiz: And that's — yeah, that's the thing I keep not being able to answer. Even if Schulman is accidentally right, even if the administrative layer really is thinning out — he himself noted that specialized roles like Total Rewards showed resilience. So it's not uniform pressure, it's — wait, no — it's selective pressure that got mislabeled as a function-wide crisis. But then the actual open question is: does any of that saved headcount get reinvested? Like, do you take the budget from the three junior coordinators whose jobs TriNet's platform just absorbed, and you hire one senior People Analytics person? Or does it just... disappear into the efficiency line on a spreadsheet?
Tess Hollis: That's the part I genuinely can't resolve. Nothing in the data — not Schulman's post, not Bersin's framing, not what Gartner published — actually tells us which of those happened. We know the ratio moved. We don't know where the savings went.
Felix Ortiz: Six likes on the founding document, and the most important question it raises doesn't have an answer yet. That's a weird place to land — but it's probably the honest one.
