HR Tech and AI Work Notes #16
The Measurers
In this past week: Cloudflare CEO Matthew Prince told Fortune on Thursday that AI had made an entire category of his workforce obsolete, and he gave that category a name. He called them “measurers.” His definition: middle management, finance, legal, internal auditing, revenue recognition. The same week, Intuit announced 3,000 cuts while reporting double-digit revenue growth, Meta’s previously announced cuts officially began, Colorado quietly gutted the AI law that was supposed to be the national compliance template, and Gallup released data showing the workers most worried about AI displacement work in exactly the functions Prince named.
And Prince was not the only CEO on a Fortune stage this week telling a version of this story. Two days earlier at Fortune's Workforce Innovation Summit, Bolt CEO Ryan Breslow said he fired his entire HR team because they were "creating problems that didn't exist" and that "those problems disappeared when I let them go." Bolt is down to about 100 employees from a peak near 800, with a 30% cut as recently as April. This isn’t about HR creating problems that don’t exist - this is about his company tanking and him looking for a boogeyman to blame as he tries to salvage what’s left.
—
Same week, same circuit, two CEOs naming the same kinds of functions as the ones that should go. I’ll be blunt about how I feel about this. Both arguments are short-sighted margin moves dressed up as strategic transformation. HR, finance, legal, and audit are the functions that quietly catch problems before they become disasters, and the value they add is mostly invisible until it is not there. Cut them aggressively and you get a few quarters of better margins, and then you discover what those teams were actually preventing. A botched RIF that turns into a discrimination claim. A revenue recognition mistake that triggers an embarrassing restatement that shakes investor confidence. A culture problem that leads to expensive turnover, lackluster engagement, and low performance that show up in lack of output. The Careerminds data I covered in #10 already showed nearly a third of AI-driven layoffs were being partially reversed within six months. I expect the next twelve months will produce a steady stream of follow-on stories from companies that took the Prince and Breslow argument seriously and then discovered what their “measurers” were actually doing for them.
🚀 HR Tech Watch
Cloudflare’s CEO named the cut. “Measurers” - middle management, finance, legal, internal auditing, revenue recognition.
On May 21, Fortune ran a profile of Cloudflare CEO Matthew Prince in which he gave the AI layoff argument a specificity most CEOs avoid. Prince said AI had rendered an entire category of his workforce obsolete and called them “measurers.” His own definition: middle management, finance, legal, internal auditing, and revenue recognition. People whose primary job is to verify, reconcile, and oversee, rather than to judge or create. Cloudflare’s internal AI usage grew more than 600% over three months, and the company’s quarterly revenue hit $639.8 million, up 34% year over year. The 1,100 cuts announced May 7 fell on those measurer functions across teams and geographies. Salespeople carrying revenue quotas were the only group explicitly protected.
Cloudflare cut 20% of its workforce in a quarter where revenue hit a record high. That decision needs an external story to explain it, and “AI made these jobs obsolete” is a much cleaner story than “record revenue did not justify the headcount we already had.” The “measurer” word lets Prince sell the cut as inevitable rather than chosen. It gives the move a clinical, almost neutral feel, when what actually happened is that a fundamentally healthy company decided to compress its workforce and is now reverse-engineering a vocabulary to make that decision sound like the only one available.
Why you should care:
I expect the term to spread fast because it offers other executives the same cover it gave Prince. It lets them point at a category without admitting that record performance is no longer a guarantee of stable headcount. The deeper problem is that the framing treats verification work as if it were transactional, when most of it is not. Financial close involves judgment under uncertainty and personal regulatory accountability that AI does not currently carry. Legal review involves judgment about which issues matter and which do not. Internal audit credibility depends on independence that AI structurally cannot provide. The work that looks like “checking against a rule” is usually “noticing the thing that does not fit the rule,” which is a different skill entirely.
What I would do internally before this framing arrives in your boardroom is sort your own functions into rough buckets: which work is mostly verification, which is mostly judgment, which is mostly creation. Then go back through the verification bucket and separate the genuinely transactional pieces from the judgment work that just looks transactional on the org chart. That distinction is what protects the functions worth keeping when somebody walks in with Prince’s vocabulary. The premium content section this week helps breakdown job descriptions like these into tasks that are truly exposed to disruption so you can put some thought and analysis into this before headlines start to create emotional responses across c-suites
Sources: TechCrunch · CFO Dive · TradingView (SaaS context) · NPR
⚡ Signal From the Field
Oracle and Eightfold launched an embedded hiring agent inside Oracle Recruiting Cloud. The deployment pattern is more interesting than the agent.
Eightfold and Oracle announced the general availability of an agentic interview intelligence product embedded directly inside Oracle Recruiting Cloud, available on Oracle Cloud Marketplace. The product runs structured interview workflows, scores candidate responses against role profiles, and surfaces what the vendors describe as bias-checked signals for hiring managers. I have not seen the product run, so I cannot evaluate the quality of what it produces.
What I want to flag is the deployment pattern, not the product. A couple weeks ago in #15 I covered Eightfold’s TalentForge announcement at Cultivate 2026, which positioned Eightfold as a platform for enterprises to build their own HR software. The Oracle integration runs in the opposite direction. Eightfold’s AI sits inside another vendor’s core HCM, available as a Marketplace install rather than as a separate platform decision. Same vendor, two very different go-to-market motions inside the same month.
Why you should care:
For HR tech buyers, the question I would want answered before this kind of integration shows up in your procurement queue is how the governance story differs from your existing HCM’s governance story. A native module of your core HCM inherits the audit trails, access controls, and data residency posture you already have in place. An embedded vendor partnership inherits less of that, and the data flow questions get harder. A separate platform inherits almost nothing. If your AI procurement intake form does not ask which of those three buckets a tool falls into, you are not actually evaluating governance, you are evaluating features.
Source: Eightfold
🔍 Reality Check
18% of U.S. employees think AI will eliminate their job in five years. In finance, insurance, and tech, the number is roughly 32%.
Gallup released Q1 2026 data showing 18% of U.S. employees say it is “very” or “somewhat” likely their job will be eliminated in the next five years because of automation or AI. In organizations where AI has been implemented, that rises to 23%. The industry numbers are the ones to call out. Finance: 32%. Insurance: 32%. Technology: 31%. Those three industries cluster at the top of perceived displacement risk and they overlap closely with the functions Prince named as “measurers” earlier in this issue.
I do not think the overlap is accidental. People who do verification, reconciliation, and oversight work can usually see how exposed their jobs are without an executive having to spell it out. The Gallup number gives that quiet awareness a measurement.
What it does not capture is what it feels like to be on the receiving end of the external narrative this week. Tens of thousands of finance, legal, audit, and middle-management professionals heard Prince say their work is the category AI replaces. And anyone working in HR specifically heard Breslow say their team “was creating problems that didn’t exist” and that “those problems disappeared when I let them go.”
Why you should care:
This is where the rest of HR leadership has a real opportunity to do something other than nod along. Yes, these roles will be disrupted. Yes, the work inside them will change substantially in the next three to five years. But “the work will change” is not the same as “the people doing the work should be replaced.” The opposite is closer to true. The people sitting in those roles right now hold the institutional knowledge of how decisions actually get made at your company, where the risks actually live, and which exceptions are worth catching. They are the population best positioned to do the AI-augmented version of the same role, because they already know what a good output looks like. Replacing them with newer, cheaper people who do not know what good looks like and pointing them at AI tools is not transformation. It is regression with extra steps.
The leadership move worth making is the opposite of what Breslow and Prince modeled this week. Tell your finance, legal, audit, HR, and middle-management populations directly that you see them, that you know how it sounds when other CEOs talk this way, and that your investment in their development through the AI transition is going to be larger over the next two years, not smaller. Mean it. Fund it. The retention upside is real, and the alternative is watching your best people leave for organizations that can credibly say something different. Adding the AI displacement question to your next engagement pulse is a five-minute change worth doing, but the survey is a measurement, not the work. The work is the conversation, and most of your peers will not have it.
Source: Gallup State of the Global Workplace 2026
🚨 Risk Radar
Colorado just gutted the AI law that was supposed to be the national template. The June 30 deadline is gone.
The Colorado General Assembly passed SB 26-189 in the closing days of the 2026 legislative session, and Governor Polis signed it, substantially rewriting the Colorado AI Act before it could take effect. The original law was scheduled to go live June 30, 2026, with sweeping requirements on employers using AI in consequential decisions: risk management programs, annual impact assessments, transparency notices, and a duty to use reasonable care to prevent algorithmic discrimination. SB 26-189 strips out the risk management programs, the duty of reasonable care, and the annual impact assessment, and pushes the effective date to January 1, 2027. What remains is a narrower notice-and-transparency framework with three duties: notice before use, an adverse action process with a right to human review, and three-year record retention. Colorado’s Attorney General has also said he will not enforce the new law until rulemaking is complete.
I wrote in #11 that Colorado was the national test case and that organizations should treat it as the template for what other states would follow. That was true at the time. It is not true anymore, and I think it is worth saying so directly. The most aggressive AI employment law in the country was rewritten by its own legislature one month before taking effect, with the Governor’s support. Other state legislators are watching that and drawing conclusions.
Why you should care:
The legal pressure on HR and compliance leaders to inventory AI systems, run impact assessments, and document algorithmic discrimination risk is materially lower than it was six weeks ago. That does not eliminate underlying liability under Title VII, ADA, and state employment law, which the EEOC continues to enforce on a disparate-impact theory whether or not the federal AI guidance is in place. It does mean that compliance teams who were standing up Colorado-specific programs can scale back. What I am more interested in is what happens to the voluntary AI governance work many organizations were doing. AI bias testing, vendor reviews, intake checklists, all of that work was about to become mandatory. Now it is optional again. Whether to keep doing it is a real decision that should be made on purpose.
Sources: Littler · Proskauer Law and the Workplace
🔢 One Number Worth Remembering
32%
That is the share of U.S. workers in finance and insurance who say it is “very” or “somewhat” likely their job will be eliminated by AI in the next five years, per Gallup’s Q1 2026 data. Tech workers were close behind at 31%. Those are the same three industries where the “measurer” functions Prince named tend to be largest. Your finance, legal, audit, and middle-management populations may already believe their jobs are exposed. Whether you have asked them, and whether your retention strategy reflects what they would say if you did, is a different question.
Premium Content for paid subscribers this week:
The 7-step Claude workflow you can test on some of your workforce job descriptions. Copy-paste prompts that you can leverage with Claude, Gemini, or ChatGPT to break job descriptions down into structured tasks and skills and identify AI disruption on the horizon
The 6-category task taxonomy that holds up across roles and stays consistent under cross-function comparison
Claude prompting at scale isn’t practical - so there’s 4-layer architecture diagram for scaling this across your whole job library, with the LLM gateway, RAG layer, and vector normalization pieces named explicitly. I built this inside of KinHire - you can too
An honest read on what is a weekend build, what is a multi-week build, and which piece I have already shipped that you can repurpose
If you have been looking for a concrete way to get ahead of the “measurer” conversation before it lands on your CFO’s desk, the workflow below is what I would build. If you are not yet a paid subscriber, this is the week to consider it.