Musical accompaniment for this week's newsletter.
Here's a fun game for your next industry conference dinner: Ask five smart people whether Workday is finished.
You'll get five different answers. All of them will be confident. None of them will fully agree. And somehow, each one will tell you more about how the speaker thinks about enterprise software than it tells you about Workday.
That's what happened this week — except instead of a Vegas steakhouse feed, it was the internet.
Joe Schmidt at a16z fired the opening shot with a piece called Workday's Last Workday — a well-argued, well-researched case that HCM is “the last large enterprise software category without a serious AI-native challenger” and that the platform shift underway will eventually do to Workday what Workday did to PeopleSoft. The piece is worth your time. The historical parallel is genuinely illuminating. The investment thesis is also obvious: this is basically a16z posting a “we're funding the thing that kills this” announcement dressed up as analysis, but at least they're honest about it.
Then Thomas Otter entered the chat.
Otter, to put it mildly, does not suffer fools gladly, even if they are some a16z big wheel. He’s forgotten more about enterprise HR software than most folks in the industry have ever learned. If you get ahead of your skis, expect him to be characteristically direct and quick to drop a withering aside. Example: “The only people asking for an AI-native HRIS are people who have a Substack and an early Soho House membership.” Yeah, that’s gonna leave a bruise.
Schmidt's claim that Workday is "the most important and least loved product in enterprise software"? Anecdotal. The supporting examples? Flimsy. Otter noted pointedly that nobody likes CRM software either, and Salesforce seems to be doing okay. He also named the central paradox Schmidt's thesis never resolves: if Workday is so unloved, why does it renew at close to 100%?
The ecosystem is the moat — not the product, not the UI, not the brand. The army of implementation partners, the decade of embedded HR workflows, the compliance surface that only ever expands. And Workday was never just about the software. It was a movement. HR leaders felt empowered by it. They belonged to something. You don't build that twice with a better UI and some agents.
Josh Bersin looked at the same company and saw a system of record transforming itself into an agentic platform. It was a tad more breathless than is needed to be, but it did bring data points: $400 million in AI ARR, 25+ Illuminate features, and consumption-based pricing signed by Nike and Merck.
Anita Lettink offered arguably the sharpest contrarian take: don't try to build the next Workday at all. Build something structurally different instead. It’s a spot-on take.
Four smart people. One company. Four completely different reads.
Here's what I think is actually going on.
The SaaSpocalypse crowd keeps making the same category error. They look at Workday and see a clunky product with unhappy power users and assume the unhappiness is load-bearing. It isn't.
Enterprise software at Workday's scale isn't really a product anymore. It's infrastructure. It's compliance. It's the organizing logic of your entire HR operating model, which, as Mervyn Dinnen points out, is a lot harder to move from AI experimentation to impact than any press release will admit.
So, the real reason Workday renews at nearly 100% isn't that customers love it. It's that leaving is a multi-year, eight-figure, career-risk project that most CHROs will never greenlight. That's not a weakness. That's a moat. A deeply annoying, kind of depressing, totally real moat.
But there's a layer beneath this that Craig Hepburn has been writing about that consistently gets underweighted in this debate. His argument is simple and cuts clean: tasks are getting cheap. Pennies on the token dollar. What AI cannot replicate is judgment — the context, the nuance, the knowing what the data doesn't say. The systems that survive the next decade aren't the ones that automate the most tasks. They're the ones that own the highest-quality judgment layer.
Apply that lens to Workday, and the picture shifts. Yes, an AI-native challenger can probably replicate the task layer — the comp cycle workflows, the performance hand-offs, the ticket queues. Schmidt makes that case compellingly. What it can't replicate overnight is the judgment infrastructure Workday has accumulated over two decades: encoded HR policy, compliance logic across jurisdictions, institutional knowledge baked into every tenant configuration. That's not tasks. That's judgment. And it's a lot harder to rebuild than a form-based UI.
Does that mean Workday is safe forever? Nope. Platform shifts are real, the historical parallel is sound, and the challenger will eventually arrive. But enterprise software doesn't get replaced by better software. It gets replaced by better software at the moment the organization has the political will, the implementation capacity, and the risk tolerance to make the switch. Those three things almost never align. When they do, it's a generational event, and it takes longer than any VC timeline wants to admit.
Workday is both the exemplar of enterprise software's challenge and the reason the SaaSpocalypse discourse keeps getting it wrong. These disparate takes are all an accurate reflection of a market genuinely mid-shift. Tasks are getting automated, the judgment layer is quietly the whole game, and nobody has yet built the thing that actually moves the needle.
Stay tuned. This one's going to take a while.
What else is going on this week?
This one from George LaRocque is flying under the radar, and it shouldn't be.
When Anthropic reprices its API — and it has — every Work Tech vendor that built its AI product layer on top of Claude has a margin problem it didn't fully model. Token costs aren't a rounding error. They're increasingly the difference between a viable product and an expensive demo.
In case George wasn’t clear enough, Mark Smith put a fine point on it: seat-based pricing is structurally broken, but there’s also a governance and measurement challenge. Enterprises don't actually have the infrastructure to track what AI is doing well enough to price it on outcomes.
Pay attention. This is a quiet problem that’s about to get real loud.
When Big Tech companies cut headcount, they aren’t saying that AI has replaced the tasks these people do. They’re saying we need to fund the massive CapEx required by AI. We’re all clear on that, right?
That’s an important distinction. New X-Team research reveals companies are confident about AI, but lack the talent infrastructure to build an AI workforce. Confidence without capability is just a slide deck. X-Team's research puts numbers on the gap between what leadership is saying about AI readiness and what the actual workforce infrastructure supports.
Things could get tougher before they get easier. CapEx is exploding as the clock is ticking on making these investments pay off. And, when you factor in Thoma Bravo’s $5.1 billion haircut on Medallia, you can see a larger reckoning: the SaaS multiples of that era were a mirage, and the bill is still coming due.
TL;DR We are creating a giant debt crisis on top of the last giant debt crisis.
I wanted to highlight this one again. Something that didn’t end up in the final edit of my interview with Coray Grove, founder of WagePath: He was a youth football coach for Ryan Wingo, the Texas Longhorns wide receiver and potential first-round NFL draft choice next year. I mean, talking about prevailing wage compliance is pretty cool. But that little factoid was too awesome not to mention. Give it a listen.
Phenom acquires Plum to verify what AI can't fake: human behavior at work. As AI-generated candidate profiles get more sophisticated, behavioral assessment data becomes one of the few signals that's genuinely hard to game. Phenom is buying the verification layer. (Press Release)
Windmill raises $12 million in seed funding. The New York-based AI-native “People Context Graph” company is betting that the real problem in workforce intelligence isn't data access; it's understanding the human relationships and context that don't show up in the HRIS. (Press Release)
Clarasight raises $11.5 million Series A. The demand for real-time headcount intelligence continues to accelerate. The companies that can connect people data to business outcomes are winning the budget conversations right now. (FinSMEs)
SnapCare and connectRN join forces to provide growth-forward workforce solutions powering better care. Two healthcare staffing platforms combining to build scale in one of the most AI-resistant labor markets around — clinical staffing, where human judgment and licensing requirements put hard limits on automation. (Press Release)
Vensure Employer Solutions completes 108th acquisition. At some point, the story stops being about any individual deal and becomes about what Vensure is actually building: the largest PEO infrastructure play in the market, assembled one deal at a time. (Press Release)
NEOGOV acquires Miller Mendel, integrating eSOPH to strengthen AI-powered background investigations. NEOGOV is deepening its position in the public-sector HR market with a background investigation platform built specifically for law enforcement hiring. The public sector is one of the least-disrupted corners of HR Tech, which makes it one of the most interesting. (Press Release)
Dex raises $5.3 million in seed funding. Early-stage funding for the London startup that is building an AI talent agent that combines matchmaking, coaching, and retention tools for both candidates and employers. (FinSMEs)
Blomma raises $5 million in seed funding. The SF-based AI career coaching platform has some blue-chip angels behind it. That’s either a sign that they have something impressive or that there's a lot of money sloshing around tech circles. Or a little of both. (FinSMEs)
Microsoft tops Wall Street expectations, reports accelerating Azure growth and $37B AI run rate. The infrastructure layer is printing money. Azure growth is accelerating, and Microsoft's AI revenue run rate just crossed a number that makes every other vendor's AI ARR claim look like a rounding error by comparison. (GeekWire)
Amazon unveils agentic AI hiring system "Amazon Connect Talent," signaling a shift toward AI-orchestrated recruitment. Amazon building an agentic voice interview system designed to handle 250,000 seasonal hires with zero human intervention. This isn't a feature — it's Amazon deciding a significant chunk of the recruiting workflow is now a solved infrastructure problem. ATS vendors: you are on notice. (Press Release)
ADP forecasts FY 2026 adjusted EPS growth of 10–11% as it raises revenue growth outlook to 6–7%. A boring infrastructure business with sticky customers is a perfectly fine place to be during a hype cycle. (Seeking Alpha)
TriNet non-GAAP EPS of $2.48 beats by $0.62, revenue of $1.2 billion beats by $876 million. The mid-market HR outsourcing model is holding up better than the doom narrative suggests. (Seeking Alpha)
Sage launches Sage HCM to connect HR, payroll, and finance for mid-market. The mid-market is where the legacy consolidation opportunity is the largest remaining greenfield opportunity in Work Tech. (Press Release)
Executive moves: Larry Dunivan — veteran of Namely, Mineral, and a string of successful HR tech exits — settles into the Executive Chair seat at HRSoft. HireRight appoints Ryan Lowe as Chief Financial Officer. Patrick Lannon joins Deel as Head of Partnerships. Atlassian's CPO takes on the AI Enablement Officer role.
What's up with AI and HR Tech litigation? The legal surface area around AI in HR is expanding fast. Heather Bussing offers a useful overview of where the case law is developing and where it isn't yet. (HR Examiner)
In the age of AI, where are new grads headed? Salesforce may offer splashy headlines about hiring new grads, but which industries are actually hiring, what roles are being created, and where the entry-level market is quietly shifting? (HR Executive)
Why the AI job apocalypse (probably) won't happen. Ezra Klein takes the hyperventilating consensus, sits it down, and walks through the actual evidence. (The New York Times)
EmpowerHR 2026: Heather McGowan on leading in an AI-driven world. McGowan's core argument is that adaptability is now the primary human competitive advantage. The Betterworks recap is a good entry point if you haven't encountered her work. (LinkedIn)
"The most bipartisan issue since beer": opposition to AI data centers. Turns out nobody wants a 500-megawatt power draw in their backyard. Hey AI: Way to bring America together. (The New York Times)
Why AI startup offices in NYC are flashy but mostly empty. The aesthetic of AI ambition collides with the operational reality of companies that run on 40 people and a lot of agents. (Wall Street Journal)
Your passwords are probably screwed. Why credential security is worse than most people think and getting worse. (The New York Times)
UKG launches into Google Cloud's Gemini Enterprise Agent Gallery. UKG's People Assist agent is now available inside the Gemini Enterprise ecosystem — meaning HR service delivery requests can be handled conversationally without leaving the tools employees already use. (Press Release)
BambooHR partners with Turn to deliver native background checks. Background screening moving natively into the HCM workflow rather than existing as a separate vendor relationship. (Press Release)
Checkr launches Profiles, giving people a verified identity that travels with them. The worker-owned credential model has been talked about for years; Checkr is actually building it. (Press Release)
Workday Government unveils Personnel Action Request (PAR) agent to modernize federal HR. The PAR workflow is one of the most painful, paper-heavy processes in government HR. Automating it is genuinely useful and a smart move to protect territory where no AI-native challenger has yet shown up. (Press Release)
Workhuman launches FutureLeaders. The recognition platform is expanding into leadership development. It’s a logical move for a company that has built significant data on how employees are actually recognized and valued across organizations. (LinkedIn)
HireClix launches JobFlow AEO. Answer Engine Optimization for job postings. Neil Costa’s team is building for the world where candidates find jobs through AI-generated search summaries rather than traditional job boards. (Press Release)
Nomad Health evolves from staffing agency to AI software company. This is a bet that the infrastructure built for healthcare staffing is more valuable as software than as a service business. (Press Release)
Klaar launches new compensation module powered by comprehensive data. This is a logical adjacency for a platform that already sits in the performance review workflow. (Press Release)
Multiplier launches global payroll payments. The race to own the end-to-end global employment stack is intensifying. (Press Release)
Jason Averbook figured out why AI adoption is broken. The problem isn't the technology; it's that most organizations are trying to layer AI onto broken processes. (Substack)
Employer branding: who controls it in the age of AI search? The answer, increasingly, is: not you. (HR Executive)
How AI is exposing the structural problem with contingent workforce software. Even if you don’t care about this category, the business dynamics that Christopher Brower talks about are pretty universal. (LinkedIn)
A conversation with a health-care-provider support bot. Apparently, the literary elite is now reading enterprise software contracts. (The New Yorker)
What the 1920s can teach us about surviving the AI revolution. The last time a general-purpose technology rewired the economy this fast, it took two decades, a Depression, and a World War to sort out the distribution of gains. So, this is, you know, the optimistic framing. (Wall Street Journal)
Understanding candidate fraud: beyond the hype. Matt Charney with a typically clear-eyed breakdown of what's actually happening with AI-assisted candidate fraud — what's real, what's inflated, and what vendors are using as a sales pitch. (Snark Attack)
This week in epic copywriting fails. Something in Zendesk's latest moves warranted a genuine “who the F approved this” reaction. See if you can catch it. The comments are worth reading too. (LinkedIn)