tldr;

• Workday Candidate Profile lets job seekers save their info once and prefill about 80% of an application at any employer on Workday Recruiting. Good feature. Overdue, honestly.

• The profile exists before the candidate picks an employer. So it doesn't live in your tenant. That likely makes Workday a controller of candidate data, not just your processor.

• Workday hasn't said publicly how retention, deletion, consent, or your DPA handle that. One analyst outlet reported it ships on by default.

• Below: five questions to send your account team before it lands on your career site.

Several Risings ago (after 8 they start running together), I found out the next action button was coming to the candidate grid. GA that day.

So I stopped right there on the floor, configured it, closed the ticket, and waited.

The recruiters found it almost immediately. They loved it. I felt like a hero. (No change request. No steering committee. Just a happy recruiter and a closed ticket.)

That's how I've hunted at Rising ever since. Not for the roadmap. For one thing I can turn on this week that saves somebody's day.

This year I thought I'd found it again.

At this year's keynote, Gerrit Kazmaier said it out loud: People roast the Workday candidate login. Constantly.

The room laughed. Then the room clapped. (Seventeen thousand people, a healthy percentage of whom have reset a candidate password on someone else's behalf.)

Then he showed Workday Candidate Profile. Save your info once. Reuse it at any company that hires through Workday. Email, one-time code, done. Roughly 80% of the application filled in before you type a word.

I clapped too. My first reaction: roll this out as soon as humanly possible. That's my next action button.

My second reaction was a question. And the longer I hung with it, the less I wanted to be first in line.

Where does the profile live before the candidate picks you?

Nobody's tenant

The candidate signs in with an email and a one-time code. No employer account. No tenant login. They build the profile, then they choose where to apply.

Which means at the moment it's created, the profile doesn't belong to any customer. It's not in your tenant. It's not in mine.

It's Workday's.

That sounds like a nitpick. It isn't. For as long as Workday Recruiting has existed, its privacy model has rested on one line: your company owns the candidate relationship, and Workday holds the data on your behalf.

Candidate Profile steps over that line. Politely. With great UX. But it steps over it.

The line you've been standing on

Workday's own privacy statement draws it clearly. When you use a Workday app your employer provides, the employer is the controller. Workday is the processor, working on the employer's instructions.

Every applicant privacy notice you've written assumes that. Every DPA your legal team negotiated assumes that. Your retention calendar assumes that.

A profile a job seeker creates directly, before any employer is in the picture, doesn't fit that model. Somebody decides why and how that data gets processed before it ever reaches you. That somebody is a controller. And it isn't you.

Has Workday said so? No. Has Workday said otherwise? Also no. The press release covers prefill, speed, and abandonment rates. Data custody doesn't come up. (Read it. It's not in there.)

I'm not the only one who noticed. HR Tech Edge pointed out that the announcement skips most of the retention, deletion, and privacy-management details.

Two copies, two clocks

Once the candidate applies, you get a copy. That copy is yours. Your retention rules, your purge jobs, your privacy process.

The profile stays where it was.

So picture the request every EU recruiting team will eventually get: delete everything you have on me. You run your purge. It works. Your copy is gone.

The profile isn't. You never owned it, so you can't delete it.

I think that's exactly how it should work. When an employer deletes your application, your LinkedIn profile doesn't vanish either. But your process, your notice, and your answer to that candidate now need to know there are two doors. Today they assume one.

Defaults someone else picked

Constellation Research reported that Candidate Profile is on by default and included in Workday Recruiting subscriptions, with GA in December 2026. Workday's own release says early calendar 2027. The date gap is probably just a slip. "On by default" is the part I'd confirm.

Because on by default means a sign-in option governed by someone else's privacy terms shows up on your career site, unless someone on your team does something about it.

Which is the irony. I came home wanting to turn it on. If Constellation is right, I won't have to.

Two more things from Rising week:

• The keynote said the profile is also reachable through ChatGPT. That's another party in the chain.

• In the Talent Acquisition breakout, unified consent management across the TA suite was on the roadmap. Marked "Future." Candidate Profile ships before it does. And that slide was under safe harbor, so the consent piece is a direction, not a commitment. The profile has a ship date. The consent framework has a disclaimer.

None of that is scandalous. It's sequencing. But sequencing is exactly where practitioners get burned, because the feature ships on a date and the governance ships "later."

It's a network, not a form

Read Workday's blog post on Candidate Profile and one line jumps out. Over time, the profile can help build a "deeper, more trusted picture" of a candidate's experience and career history.

Prefill is the feature. The picture is the strategy.

A profile that spans thousands of employers, gets verified, and gets richer with every application is a candidate graph. Workday has never had one of those. You know who has? LinkedIn.

I'm not saying that's sinister. I'm saying it's a much bigger product than autofill, and your governance should be sized for the bigger thing.

If you read my Silver Lake piece, you know the Monday action I keep coming back to: inventory what you own versus what you rent. Your candidate data just picked up a new line on that inventory.

In fairness

Candidate login hell is real. Workday cites a 60% application abandonment rate from length and friction (I believe it's higher), and anyone who has supported a career site has worked the "I can't log in" ticket more times than they'd like to admit. Fixing that is good for candidates, good for recruiters, and good for the reputation of everyone who configures this stuff.

The design is restrained in the right places, too. It doesn't rank, recommend, or reject anyone. Candidates choose what each employer sees. And a verified, reusable identity is a real brick in the wall against the bot-vs-bot fraud mess. (A profile that persists is a lot harder to fake than an application that doesn't.)

And Workday's privacy people are good at this. I'd bet the controller question has an answer in a contract somewhere.

The problem isn't that the answer doesn't exist. It's that you need it before your career site changes, not after.

What to do by Next Monday

Five questions. Send them to your account team and copy your privacy office.

1. Who is the controller of Candidate Profile data, before and after a candidate applies?

2. Is it on by default? Can we turn it off, and can we turn it off by region?

3. When we purge a candidate, what happens to the profile? Who handles the deletion request?

4. What changes in our DPA, if anything? Is an addendum coming?

5. When a candidate uses the profile through ChatGPT, who's the processor on that path?

Next, the internal homework. Flag your applicant privacy notice for review. Add a row to the retention calendar. Decide who owns the toggle before Workday decides for you.

I still want this feature, the earlier the better. I'm just not flipping it on until those five answers come back.

Some years ago, the hero move was configuring something on the floor before lunch. This year, it's an email.

Your candidates were yours because nobody else could reach them.

That was never a policy. It was the architecture.

And the architecture just changed.

— Mike

Director HR Technology | (Allegedly) Not a Data Controller

P.S. — I asked this in a LinkedIn comment over the weekend. It got four likes and a "great point!" Which, in LinkedIn economics, is a standing ovation. If you're on an Global, EU, or UK recruiting team and your privacy office is already asking about this, hit reply. I want to know what they're asking.

tldr;

Rising is over. The streamers got cleaned up. Your configuration didn't.

Workday's keynote line of the week: an agent not grounded in business rules is lawless. True. Also: the business rules are your configuration, and somebody wrote them in twenty minutes.

Lawful and accountable are different things. Lawful AI executes a policy decision perfectly and at scale, and leaves the person affected with nobody to call.

Three things I watched in Las Vegas, each one a guardrail test: a performance review that arrived pre-rated 4.5, a recruiter who shows up after the shortlist, and an agent that runs on your permissions.

Four guardrails, one line for a ticket, and the scoreboard of what you can actually turn on when your CHRO comes back asking.

Somewhere in the back half of the Kelly Clarkson set on Wednesday night, a guy near the front launched a few streamers into the crowd. Long ones. Industrial grade. Kelly Clarkson stopped mid-set to talk to him about it. She didn't know how he got those in, she said, but it was really cool. Which is the most generous security review anybody received all week.

He's gone now. The streamers were swept up before the house lights came on. Rising is a memory, the badges are in the trash (or hanging in an office), and thousands of us are back in tenants that look exactly like they did on Sunday.

Which is the point. The streamers got cleaned up. The configuration didn't.

On Tuesday morning Aneel Bhusri had stood on the keynote stage and said an agent that's not grounded in business rules might move fast, but it is essentially lawless. An agent grounded in the rules is lawful. The room liked it. I liked it.

I also knew who writes the rules.

Years ago, I built an auto-approval step into a business process. Requests under a threshold skipped human review, because the review was a bottleneck and everyone was tired of it. Took twenty minutes. I logged it as a config change.

Calling it a config change was generous. It was a policy decision about when a manager's judgment stops mattering. Nobody voted on it. Nobody debated it. It went in as a task on a sprint board, and I was the only person in the building who fully understood what it did.

That's the job. That's always been the job. What I watched last week is Workday putting that job on rails and handing it a throttle.

Lawful versus accountable

The last two posts were about other people's incentives, and all of that holds. But in talent acquisition, AI is a user problem. In HR technology, we're the plumbers.

So take the keynote line seriously, because it is the best thing Workday has said about agents and it is also a problem. Lawful means the agent follows the rules. It doesn't ask whether the rules are right. The rules are the configuration: every security group that says who deserves to see what, every condition rule that decides which humans are worth stopping for, every auto-advance that says a human's judgment isn't needed here. None of those arrived as policy questions. They arrived as tickets.

Lawful AI executes that ticket perfectly, ten thousand times a month. That's compliance with the tenant. The person on the other end is still waiting for someone accountable.

The Sunday before the keynote I sat at a table with twelve practitioners, mostly midsize shops, at the pre-Rising Share-a-thon. Three of the top twelve had signed the UMSA. Zero had an agent in production. Zero had a governance structure for an agent that goes sideways. When I asked who catches it, the best answer in the room was "hopefully Compensation."

Twelve. Three. Zero. Zero. Keep those numbers in your pocket for the next three sections.

And that table was typical. Most of the responsible-AI conversation I heard all week, on stages and over drinks, was being had by people who haven't activated one in prod yet. Nobody should be embarrassed by that. It's the finding.

Exhibit A: the review that arrived pre-rated

Sana Coach, early access as of last week. The pitch is reasonable: managers walk into feedback conversations unprepared, so let the agent assemble the signals.

The demo showed a Q1 performance review labeled "Prepared by Sana AI." The slide behind the presenter listed the inputs: GitHub, Jira, Confluence, and Slack, plus peer and self assessments. Overall rating, 4.5 out of 5, already filled in. Submit button, top right.

(Workday's own blog describes Coach more modestly, as drafting from goals, check-ins, feedback, and development activity. I'm reporting the slide I photographed.)

Here is the governance question: When that 4.5 feeds a merit cycle, who signed it? Which human's judgment does that number represent? "Prepared by Sana AI" is not a person. The manager who clicked Submit will say the system drafted it. The system will say the manager submitted it.

And notice what the agent measured: what was instrumented. Commits, tickets, wiki edits, chat. A Marketing Coordinator evaluated on her Jira hygiene. If the agent writes the review, whose judgment is the employee being evaluated on?

Exhibit B: the recruiter who shows up after

The Talent Acquisition Agent went to early adopters on Oct 13, GA planned for early 2027. The agent flow in the breakout went like this: strong candidate surfaced, application reviewed, fit explained, interview scheduled, hiring team aligned. Then the recruiter contacts the candidate and starts building rapport.

Credit where due. Workday's release says outreach still requires recruiter approval, and Kazmaier said on the record that Workday does not automate hiring decisions. I believe both. But look at where the human enters the sequence. After the review. After the fit explanation. After the scheduling.

When the human shows up after the decision, the loop has already closed. They're standing at the end of it.

One more thing from that room, and I'm raising it as a question, because it was a demo and demos lie. The surfaced-candidate card for "Eva" had her photo on it, at the stage where a recruiter decides who to look at. Maybe that's the product. Maybe it's a mock-up. Either way, what appears on the card at the moment of first impression is a configuration decision somebody makes, and it is exactly the kind of decision that never gets a policy conversation.

Exhibit C: the agent that runs on your permissions

Wednesday, the Extend session. Workday declared Developer Agent, Custom Agents, and Agent-Ready Tools generally available in what is now called Extend AI. You can build apps and agents in plain language from Claude Code, Codex, or Cursor.

The presenter typed one paragraph: build me an app for managers to nominate directs for internal mobility, show tenure, promotions in the last three years, compa ratio, kick off a BP to the manager's manager. The agent generated the pages, the WQL, a security domain called ManageNominations, a business object, and the business process. His number for the whole build: about four hours and ten minutes, including debugging. (The debugging was visible. The approver's inbox still had a "Business Process Definition Error" from the day before.)

I don't say that to dunk. Four hours to a working app in a dev tenant is remarkable. Build time collapsed. Governance time stayed exactly where it was. Security review, testing, and change control aren't in that four hours, and the agent wrote its own security domain over comp data. Your job just became reading it.

Then the part that matters most. The generated agent manifest said it recommends roles "with the worker's permission." That phrase is a sentence in a prompt. A model will usually honor it. A business process step enforces it. If your guardrail is a sentence in the prompt, it's a suggestion.

And in the Agent System of Record, the configure screen showed Execution Mode: Delegate. Workday's documentation defines it plainly: the agent performs tasks on behalf of a human and uses the worker's permissions.

Every over-provisioned security group, every stale role assignment, every "we'll clean that up after go-live" from 2019 now has an agent attached to it, moving at agent speed. Your security audit just became your AI readiness project.

The four guardrails, after Rising

One. Produce the list of decisions your system makes without a human. What it decides, as distinct from what it recommends. Auto-advance, auto-route, auto-approve, auto-reject, auto-flag, and now any agent running in Ambient mode, which the docs define as the agent performing tasks as itself. Most organizations cannot produce this list on request. The twelve at my table couldn't. Write it calmly on a Tuesday, because the version you write during an incident is worse.

Two. Ask who the worker is when an agent acts. Your business process framework assumes a human actor. Initiated by. Approved by. Delegation. Workday gave a technically complete answer last week: in Delegate mode the worker is you, and in Ambient mode it's the agent, with its own security group and a clean audit trail.

A good security admin will tell me that solves it. Transaction history shows the bot in Updated By. Traceability, done. And technically they're right, which is exactly where the argument gets missed. Logging a system account satisfies the database. When an EEOC inquiry asks which human authorized this disposition, "ISU_AI_Recruiter_Bot" is a technically complete answer and a legally useless one. Integrations move data. Agents exercise policy authority, and we have no framework concept for that.

So ask your vendor in writing, and skip "is it logged." Ask which human holds the authority the agent is exercising, and where that's recorded. Watch how specific the answer gets.

Three. Reversibility beats prevention. Some automated outcomes will be bad no matter how much you prevent. What you can guarantee is that every one is detectable, explainable, and undoable. "We need governance budget" gets you a nod. "We currently cannot undo this at scale" gets you a follow-up meeting.

Four. Somebody has to be able to answer the person. Why wasn't I advanced. Why was I rated 4.5. Why did I get flagged. If nobody can answer without saying "the system did it," you have a liability with a nice interface.

One paragraph on benefits, because the same pattern is already outside TA. In September, Workday announced Total Benefits, with benefits guidance through the Self-Service Agent expected to be GA in October. (I haven't seen the GA confirmation yet; check your tenant.) An employee asks which plan to pick and the agent answers from your eligibility rules, your plan descriptions, your life event paths. The benefits specialist who knew the dental text was three years stale is no longer standing between the config and the employee. Configuration was always policy. Now it's also speech.

The concession, and the credit

Easier said from where I sit. If you're a team of one supporting eleven modules, "inventory every automated decision" is a beautiful idea you do not have hours for. I've been the team of one. So pick the process with the highest impact, something touching pay, termination, or candidate disposition, and inventory that one.

The harder objection is standing. When a VP hands you a ticket to turn on auto-advance, you can't demand an ethics review. Correct. Ship it. And write one line in the ticket:

Automated policy delegation: Step 4 configured to auto-approve without human review. Policy owner: [name].

You still ship the config. Nothing escalates. What changes is that a policy decision now has a named owner instead of an implied one. Write it as documentation. If it reads like a shield, you've annoyed a VP for nothing.

And credit to Workday, because the plumbing is better than the keynote suggested. The Agent System of Record registers agents inactive until a human activates them. "Available To" uses the security group types you already know. The tool list is explicit (fourteen tools on the demo agent, including find performance review, which a career-advice agent doesn't need and which the builder won't trim for you). Agent Passport is in early access, GA planned by year end. And Workday signed to acquire C1.ai, an agent identity company, which tells you they know "who can press Activate" is about to be one of the most powerful questions in the tenant.

Opening the doors is right, as long as somebody's still checking IDs. The somebody is you.

The scoreboard, for the Monday your CHRO comes back

Because they will come back asking why you don't have this yet.

Available now: Sana as the new Workday Home (admins choose when to enable it). Developer Agent, Custom Agents, Agent-Ready Tools in Extend AI. Decision Intelligence. Data Cloud. Agent System of Record (GA since February).

Early access: Talent Acquisition Agent autopilot (GA planned early 2027). Sana Coach. Sana Concierge (GA expected Q1 2027). Sana in Claude (GA expected late 2026). Agent Passport (GA planned year end).

Early 2027: Workday Candidate Profile (prefills roughly 80% of the application; Workday says it doesn't rank, recommend, or reject). Payroll, Accounting, and Deployment Agent autopilot, early adopters.

Still just a number: 17 agents in production, by Workday's count. Production at Workday means something different from production in your tenant.

Most of the HR-facing agents are early access. The platform underneath them already shipped. And the governance questions live somewhere other than the roadmap: in the automated decisions already in your tenant that nobody has reviewed since the day they were configured.

Where this all lands

TA got the vivid version because the arms race is loud. TA is the preview. The point is everything after it. The same pattern is in performance now, in benefits this quarter, and it's coming for absence and comp. Every process where a model can score a human faster than a human can review the score.

In TA the accountability is diffuse. In HR technology it has a name, an employee ID, and a security group.

Responsible AI in this profession is a habit: assume you're accountable for what you make possible, and for what you decide.

Bhusri is right. An agent grounded in the rules is lawful. I still think about that auto-approval step, because it was lawful too. It followed the rules perfectly. I wrote them.

The guy with the streamers is gone. That step is still running.

Twenty minutes. Logged as a config change.

Disclosures: I spoke last week on a Rising-adjacent panel hosted by a company that sells agentic workflow software; no product names appear here and none were requested. Workday announced a partnership with Anthropic on Oct 13; Claude helped me fact-check this issue, so weigh any Anthropic-adjacent analysis accordingly.

— Mike

Director HR Tech | Wrote the Laws on a Tuesday

P.S. Justin's 3D Lab ran the capture infrastructure for Rising: one QR tag on my backpack strap, an embossed logo on the back. Final tally: 91 scans. Six new subscribers during the week. And no way to attribute the six to the tag, because the UTM configuration was mine, I did it in about twenty minutes, and I'm told it was a config change. Unless there's an argument from the community, he's getting the win. Ninety-one people scanned a tag a twelve-year-old printed and I didn't see a single one of them do it, which is the most honest description of an automated process I heard all week.

P.P.S. Spreading Rising across three casinos was a lot. I spent most of the week shuffling between Caesars Forum and the Venetian, and one day I Ubered down the Strip for a day at Resorts World. San Diego is one convention center, and I'm holding them to it. Happy to report the titanium held up. My ego did not.

tldr;

• California signed three workplace AI bills on September 30. One of them got effectively all the coverage.

• That one, SB 947, is also the furthest out: July 1, 2027.

• Two others go operative January 1. And a fourth bill, from last year, turns into a hard filing deadline on May 12.

• All four ask your tenant the same question: can you prove which system produced this output, what data fed it, and who looked at it before you acted.

• This is a lineage problem, and you either have the log or you don’t.

Newsom signed the No Robo Bosses Act on September 30, and by Wednesday I had seen the same headline about eleven times.

AI can’t fire you in California anymore. Humans required. Landmark, first in the nation, etc.

All true. Also effective July 1, 2027, which means the most-covered workplace AI law in the country is the one you have the most time to think about. The news cycle has excellent instincts for drama and none whatsoever for your project calendar.

It is also, and I say this with affection, called the No Robo Bosses Act. A real statute, with a real penalty schedule, named like a Saturday morning cartoon. I’ve spent two years arguing that vendor branding has a shelf life of about nine months while the architecture is permanent. Legislative branding works the same way, and it is just as good at deciding which of your problems gets read about.

He signed three bills touching workplace AI that day. Two of the others go operative January 1, which is three months out. And there is a fourth, from last year, that never entered this news cycle at all, and it comes due on May 12.

That one first, because it is the one that will actually eat your year.

SB 464 is a job catalog project wearing a compliance costume

California pay data reporting currently buckets your workforce into ten EEO-1 job categories. SB 464 replaces that with twenty-three categories built on the SOC system.

Read that again with your “I own the job catalog” hat on.

That is a remap of every job profile you have against a classification scheme you probably don’t currently store. It’s being told to sort the laundry into lights and darks for eight years, and then being told this year it goes by fabric weight. Ten to twenty-three isn’t a longer list of the same thing. It’s a different taxonomy, and nothing in your tenant is going to volunteer the crosswalk for you.

Here’s the part that may be ruining your Monday.

SB 464 was signed in October 2025, and employment firms covered it properly at the time. It just got filed under pay transparency rather than AI, so it surfaced in exactly zero of last week’s roundups.

The twenty-three categories attach on January 1, 2027. The first report filed under them is due May 12, 2027, and it covers the 2026 calendar year.

Count the months. The report is seven months out. The year it covers is the one you are ten months into right now.

You are not classifying going forward. You are crosswalking a year of history you already finished generating, under a taxonomy you weren’t using at the time. Nobody is going to re-interview 2026 for you.

And the enforcement half is already live. Since January 2026 the penalties are mandatory rather than discretionary: a hundred dollars per employee for the first violation, two hundred for the next. For a 500-person California headcount that is fifty thousand dollars, then a hundred. Mandatory, as in the court no longer gets to be charmed by your explanation.

SB 951 asks a question you cannot answer retroactively

SB 951 amends Cal-WARN, effective January 1. If a mass layoff or relocation is caused “in whole or in substantial part” by an AI system or other automated technology replacing positions, your notice now has to include four new things.

The count of affected positions attributable to the automation. The specific job functions being automated. The category or type of AI system responsible. And a line at the top of the notice that reads, in plain language, “This notice is for a technology displacement.”

The legislature wrote that sentence out for you. Verbatim. In the statute. That is roughly how much faith they had that anyone would volunteer it.

But check out the third requirement, because that’s the one with teeth.

Somebody has to name which system automated which roles. Not in the abstract. On a legal notice, with a date on it. Which means the attribution was tracked while the automation was happening, by someone, in a form you can retrieve.

Nobody tracks that today. Automation arrives as four initiatives with four business cases and zero shared ledger, and eighteen months later a VP says the word “efficiency” and the headcount moves.

The attribution is knowable. It’s just only knowable while it’s happening. You cannot write the receipt after the meal.

The statute also doesn’t define “in substantial part,” or “other automated technology,” or what to do about a reduction with mixed causes. So you get to make that call. Which is a polite way of saying someone will make it for you at two in the afternoon on the day the notice is due.

AB 1883 is an inventory question

AB 1883, also January 1, bars employers from using AI-based surveillance tools to recognize, infer, or predict an employee’s emotional state, or to collect neural data.

Five hundred dollars a violation. Which reads like a parking ticket right up until you notice the statute is not counting incidents.

The neural data half is science fiction for most of us. The emotional state half is not.

Go look at what in your stack scores sentiment. Engagement survey free text. Anything doing tone analysis on written feedback. Anything a vendor sold you by using the word “pulse.” I’m not telling you any of that is illegal, because the answer depends on specifics well above my pay grade and well outside my license.

I’m telling you that in three months somebody is going to ask you for the list, and the list does not currently exist.

And now the famous one

Which brings us back to SB 947, where the actual practitioner payload is buried in a clause nobody quoted.

You can’t rely solely on an automated decision system for discipline or termination. Fine, obvious, nobody was going to admit to doing that anyway.

But when an ADS is the primary basis, a human has to review and corroborate the decision “using the information that produced the output.” And if that corroboration turns up anything inaccurate, incomplete, or misleading, you can’t use the output at all.

Not glance at the score. Review the inputs that made the score.

That is a lineage requirement with a human attached. You need to hand a reviewer the underlying data, keep a record that they looked at it, and have that record survive a Labor Commissioner asking about it two years later.

Nine months to build that, document it, and train the people who execute it. If you have ever changed a business process that touches discipline, you know nine months is one release cycle and a prayer.

Which, sure. Except I already wrote the piece about how that goes.

Back in August I wrote about twenty people approving something by not questioning it. The approval chain existed. Every signature was real. Nobody read anything. The governance was architecturally perfect and functionally decorative.

California is about to legally require exactly that mechanism, and I have very little confidence it will work better because a statute asked nicely.

Where the other side has a point

I don’t want cynicism to read as an argument against the law, because it isn’t one.

A human in the loop, even a bored one, beats no human. The notice requirement is genuinely good: telling someone an automated system helped end their job is the bare minimum of treating them like a person. And the lineage work is work you should be doing anyway. The state didn’t invent the obligation. It put a date on one you’d been deferring.

That’s usually how this goes. The regulation isn’t the burden. The regulation is the invoice for eight years of nobody asking.

Monday

Three things, none of which require a lawyer.

Find out who owns the SOC crosswalk. Not eventually. This week. It’s a reconstruction project with a seven-month clock, and those always run long. At most places it is currently nobody’s job.

Inventory every system in your stack that produces a score, a classification, or a recommendation about a person. Not AI systems. Scoring systems. The statutory definition is broader than your mental model of it, and this one inventory feeds three of the four bills.

Map the data and the human for each one. What feeds it, who reviews the output today. Where the answer is “nobody,” that’s your January gap list.

In eight days I’ll be in Vegas watching Workday spend four days selling agentic HR to forty thousand people. California just spent one afternoon writing rules about agents. As far as I can tell, nobody on either stage is talking to the other one.

The vendors will sell you the agent. The state will ask you who approved its output.

Only one of those conversations happens inside your tenant.

– Mike.

P.S. SB 947 got the coverage because it had the best name. SB 464 got filed under the wrong category for a year and is going to cost you more hours than the other three combined. There is a lesson in there about taxonomies.

Sources

Dates and penalty figures in this piece are drawn from the following. Nothing here is legal advice. If any of this touches your tenant, your employment counsel is the person to call. And don't let them tell you I sent them to you. I didn't.

SB 947, the No Robo Bosses Act. Signed September 30, 2026. Effective July 1, 2027.

• Ogletree Deakins, California Governor Signs 3 Bills Targeting AI and Workplace Surveillance

• Mondaq, California Passes No Robo Bosses Act

• Crowell & Moring, New Guardrails on Automated Decision Systems in Discipline and Termination

• Legal 500, California Passes No Robo Bosses Act

• CNBC, Newsom bans AI “robo bosses,” reversing his earlier veto

• Office of Sen. McNerney, signing announcement

SB 464, pay data reporting. Signed October 13, 2025. SOC categories attach January 1, 2027. First report due May 12, 2027, covering calendar year 2026.

• Atkinson, Andelson, Loya, Ruud & Romo, California Revamps Pay Data Reporting Obligations

• Seyfarth Shaw, Preparing Now for California’s Updated Pay Data Reporting Requirements

• Seyfarth Shaw, California Amends Its Pay Data Reporting Requirements

• Nilan Johnson Lewis, What Employers Need to Know Before 2027

• Payne & Fears, California Expands Pay Data Reporting Obligations Under SB 464

• Liebert Cassidy Whitmore, SB 464 Expands Pay Data Reporting Requirements

SB 951, Cal-WARN technology displacement notice. Effective January 1, 2027.

• Littler, California Amends its WARN Act (Again)

• California Legislature, SB-951 bill text

AB 1883, workplace surveillance and emotional-state inference. Effective January 1, 2027.

• Mondaq, California Passes AB-1883 Targeting AI Workplace Surveillance of Emotional State and Neural Data

• California Assembly, committee analysis of AB 1883

Effective-date validation

• CalChamber HRWatchdog, New 2027 Employment Laws

tldr;

• My best-performing post ever reached 71,600 people and converted six subscribers. One user group room of about 170 converted forty-five. Per person, the room outperformed the feed by roughly 3,000x.

• Reach was never the scarce resource. Trust is. Feeds manufacture awareness; rooms manufacture trust; only trust converts.

• The RUG is not the farm league of this ecosystem. Per person, per minute, it’s the majors.

• Three moves, whether you run a room or just stand in one: make every meeting a capture event, give every speaker the recap promise, and trade speakers between rooms.

• I’m saying all of this to a room full of RUG leaders at Rising next Monday. Consider this the leaked memo.

On July 6th I posted the best-performing thing I have ever put on LinkedIn.

71,600 impressions. 173 engagements. My follower count jumped. My phone buzzed for two days. By every metric the platform shows you in big friendly numbers, it was the peak of a year of writing.

And that week, my newsletter (the thing I actually care about, the thing all of this allegedly feeds) gained six subscribers.

Six.

Now the other number. Two months earlier I stood in a room at the TOLA regional user group and gave a twenty-minute talk to about 170 practitioners. Forty-five of them subscribed.

Do the math with me, because I resisted doing it for weeks. The feed converted at eight thousandths of one percent. The room converted at roughly one in four. Per person reached, the room outperformed my biggest “viral” moment by a factor of about three thousand.

Three thousand. That is not a rounding disagreement. That’s two different universes pretending to be the same channel.

Reach was never the scarce resource

Here’s what those two numbers finally forced me to understand.

Feeds manufacture awareness. Rooms manufacture trust. And in this ecosystem, only trust converts, because of who we are and what we’ve survived.

Think about who this audience actually is. People who’ve sat through a decade of vendor demos that didn’t survive contact with their tenant. People who can smell a sponsored take from the parking lot. People drowning in analysts and vendors and absolutely starving for operators, for one voice that has personally loaded the EIB and lived with the consequences.

An audience like that does not subscribe because a post crossed their feed at a lucky hour. They subscribe when they’ve decided you’re real. And the deciding happens at human bandwidth: a talk, a hallway question, a conversation where you said the thing they were already thinking and couldn’t say out loud at work.

You can’t scroll your way to that. You have to be in the room.

The farm league fallacy

Which brings me to the thing our ecosystem gets “bass-akwards”.

We treat the conference talk as the big time and the regional user group as the farm league. The Rising Slot has hundreds of seats and production values. The RUG has forty folding chairs, a projector with one dead HDMI port, and someone heroically wrangling a sign-in sheet.

Now run the conversion math on those two rooms. The conference audience forms approximately zero trust with any individual human on that stage; that’s not what conferences are for. The forty people in the folding chairs will remember the speaker by name, connect the next morning, and tell a colleague. Per person, per minute of attention, the RUG is the most valuable stage in the entire Workday ecosystem, and it’s treated like a rec league.

The rooms were never the farm league. The rooms are the majors with bad lighting.

Three moves

So here’s what I’d do about it, whether you run one of these rooms or just stand in them. (Next Monday I get ten minutes in front of the people who run them, at the Community Summit before Rising. This is that talk, a week early, because the people in the folding chairs deserve the memo too.)

1. Make every meeting a capture event. Attention that isn’t captured within 24 hours evaporates. A QR code on the last slide. A one-page artifact people leave with. A follow-up that arrives while they still remember why they cared. If you speak and there’s no way for a convinced person to stay connected to you, you didn’t give a talk, you gave a performance.

2. Give every speaker the recap promise. The single best conversion mechanic I know, and it costs nothing: end with “I’m going to write up what this room said, anonymized, and send it out next week. Scan here if you want it.” It works because the audience co-owns the artifact. They’re not subscribing to a newsletter; they’re claiming a conversation they were part of. Room leaders: teach this to every speaker you book, and yours becomes the best gig in the ecosystem.

3. Trade. Rooms compound when leaders swap rosters. A practitioner who lands in Philly should be on a plane (or a Zoom link) to Dallas the next quarter (as a Philly guy, I can't believe I just said that out loud! Now if the Birds split this year with Dak and the ‘Boys youse can blame me). We are all running the same experiment in isolation, rebuilding the same speaker list from scratch. The moment the rooms start trading, every room gets better, and the operators this audience is starving for get a circuit instead of a one-off.

The part where I ask

If you run a room: your speaker calendar is the most valuable thing you control, and my own calendar is open. Trade with me. Trade with each other. Come find me Monday night and let’s make it concrete.

If you’ve never spoken at one: the stage you think you’re waiting for already exists, it’s twenty minutes from your house, and the person who runs it is perpetually one speaker short. You do not need permission or a keynote. You need a war story and a QR code.

The analysts have the feeds. We have the rooms.

I know which one I’d bet on. I have the spreadsheet.

— Mike

Director HR Tech | Six Subscribers Wiser

P.S. Speaking of capture infrastructure: the tag flies to Vegas next week. Designed, printed, and invoiced (with a brand premium we’ve agreed to stop discussing) by Justin’s 3D Lab. You voted on the design. Every scan is tracked, and the proprietor is aware that the number to beat is six. He has asked me to communicate that he is, quote, “not worried.”

The Department of First Things First. For the people who do the work.

tldr;

I spent the summer telling you to audit your bus factor. In September, the universe scheduled my exam and didn't let me study.

What held while I was gone had one property in common: the why was written down, close to where the confusion would happen.

What generated texts to a man on painkillers lived exclusively in my head. I could have named every one in advance.

Also in this issue: the conversion machine I built in public this summer is finished, and its deployment map is my Rising schedule.

Bring your audit findings to Vegas. Not going? Subscribing is attending remotely. Same deal as always.

Somebody replaced one of my hips two weeks ago. I'm told it went great. I was asleep for the interesting part.

One detour before I close the medical file: thank you to Debby Schroeder, Lawrence Berra, and the whole crew from my former team at Magellan Health for checking in on me, and for sending the two finest gifts in orthopedic history. A mug that reads "I just had a joint (replacement)," and a Christmas ornament for "Christmas with my new hip 2026." It's hard to feel sorry for yourself when your coffee mug is doing a bit.

That's the last you'll hear about the hip...

The part worth your time is what happened to everything I'm responsible for during the ten days I was, for the first time in thirty-plus years of working, genuinely unreachable.

Not "checking email from the lake or the shore" unreachable. Unreachable.

All summer I've been telling you to run the audit. Who else knows. Where the why lives. What breaks when the one person is gone. Then the universe looked at my calendar, picked a Wednesday in September, and scheduled the exam.

What the test actually measures

Here's the finding, and it held with more consistency than I expected.

The things that kept running weren't the well-documented things. Nobody opened a runbook. Nobody has ever opened a runbook. What existed was context sitting close to where the confusion was going to strike: a note in the field, a dated line explaining why a rule exists, a two-sentence handoff written like the reader was smart.

And the things that generated texts to a man on a couch full of painkillers? Every single one was a spot where the answer lived only in my head. I knew which ones they'd be. I'd been meaning to write them down since roughly June.

I know my failure mode because I am the failure mode. Years ago, I built a geocoding script at a company I no longer work for. Three plus years after I left, somebody there called me. Not to ask how it worked. To ask whether it was load-bearing. Nobody inside the building could answer that question, so they called a guy who'd been gone since the last administration.

On that one I wasn't the archaeologist. I was the ancient civilization.

The audit isn't a compliance exercise. It's a prediction of which texts you get on your worst week. Yours is coming: a vacation, a resignation, a jury summons, a hip. The only variable is whether you take the test on your own schedule.

One page, ten minutes. Run it while your bus factor is still hypothetical.

The Bus Factor Self-Audit

The other thing I built this summer

Now the part I've been assembling in public since July, possibly without telling you it was all one machine.

In July, the best-performing post I've ever written reached 71,600 people and converted six subscribers. Six. I'm doing the full math on that next week, because it turned into an argument I'm delivering in a room full of RUG leaders. Short version for today: reach without capture is weather. It passes over you and nothing grows.

So instead of posting harder, I spent the summer building a net.

The one-page audit, now wearing a QR code, so a talk becomes a takeaway becomes a subscription.

The tag. Designed, printed, and invoiced by Justin's 3D Lab (brand premium included, and we've agreed to stop discussing it, I think, but every time I have an update, he wants another change order). You voted on the design. It flies on my backpack in two weeks, with a small production run going to interested RUG leaders.

The landing page every scan hits, with a scoreboard attached, because the proprietor of the Lab is in active competition with that 71,600-impression post. The number to beat is six.

The recap promise. Every room I'm in at Rising ends the same way: I write up what the room said, anonymized, and it goes to subscribers. The conversation becomes the artifact.

None of this is sophisticated. That's the point. A QR code, a landing page, a kid with a printer, and a promise I intend to keep.

Steal all of it. It works at a forty-chair RUG on a Tuesday afternoon.

Where to find me

Sunday: I'm facilitating an hour-long AI governance roundtable at the pre-Rising Share-a-thon. If you read the Twenty People post and have opinions, come sit at the table and bring them.

Monday, 1:00: Ten minutes at the Workday Community Summit on why your room is gold, in a room run by Greg Higgins, in front of the people who run the rooms. It's the argument I've been circling all year.

Monday, evening: Dinner with those same people. If you organize a RUG and you're on that list, find me before the entrées.

Tuesday, 5–8: A forum on the responsible use of AI, hosted by Interlay with Ceci and the Well Built crew, at the Venetian. This is one you can actually get into: tickets are through Well Built, and their premium subscribers get first crack. If Sunday's roundtable is the version where twelve of us discuss at a table, this is the version with a bigger room.

Braindates: I made the case last week that the braindate is the product. So I should probably practice what I preached. I'm not hosting any yet. I will if anyone wants one. DM me and I'll put something on the board, because thirty scheduled minutes with a few practitioners beats a hallway collision every time.

All week: hallways, sessions, the tag on the backpack. Scan it, say hi, argue with me about your audit results. That last one is a genuine request, and it's my favorite genre of conversation (just ask Justin).

If you're not going to Vegas, you're not missing the substance, because the substance ships here. The roundtable writeup, the room argument in full, the scoreboard, and an honest accounting of what mattered versus what was booth glitter.

Subscribing is attending remotely. That's always been the deal.

Two weeks out. The machine is built, the hip is bionic, and for once the hopper is full.

See you in the hallways.

— Mike

Director HR Tech | Back in the Chair

P.S. Before the tag flew anywhere we scan-tested it in bad lighting, because a QR code that only works under a desk lamp is a prop. Justin's prints passed on every phone we tried. Mine failed. He picked my version up, turned it toward the window, flipped it over twice, and said "huh." No diagnosis. No follow-up. Just "huh," and then he went back to his own print run. I've sat through hour-long design reviews that delivered less.

P.P.S. - Hey Christian, sorry about your Commanders week one. It came down to the wire though, and they looked pretty good. Too bad Jayden went down against the Cowboys. Greg (and Kris, Karthik, Sheena, and the rest of my Dallas friends), we're coming for you. 10/26 is circled on my calendar. Go Birds 🦅.

tldr;

You could skip every session at Rising and still win the week, if you had the right ten conversations.

The sessions are the excuse to be in the building.

The un-Googleable stuff (the failure nobody documents, its 2am workaround that held, how they got Finance and HR to agree on one headcount number) lives in people’s heads. The only way to extract it is to stand next to them.

The best conversations have a little structure. A roundtable. A small room of people who do your exact job with one real question on the table. Spend your time where the ROI is higher.

I'm writing this from a recliner with an ice machine where my arthritic hip used to be. Nobody warns you at 20 that the warranty runs out earlier than you expect.

Ok. Enough whining.

Five days ago, a very competent man in a mask (thanks, Dr. Vikoren) rearranged part of my skeleton. The instructions were clear. Rest. Ice. PT.

Don't be a hero.

So naturally, I'm building my Rising conversation list.

Here’s what I figured out somewhere around my fourth Rising: the sessions are not the point. The sessions are the excuse to get everyone into the same building at the same time. What happens between them is the actual event.

The sessions are the brochure.

Everything on the main stage will be online by November. Replay exists. Release notes exist, and they’re written more clearly than any keynote, because nobody’s reading release notes to a room with a light show.

You know what’s not online? The person who leans over during the session and says “yeah, we turned that on, here’s what actually happened.”

That sentence is not in any deck. It’s in someone’s head. And the only way to get it out is to be standing close enough to hear it.

The un-Googleable premium.

Everything you can look up, you can look up from your couch. (I’m on my couch. I’m something of an expert.)

What you can’t look up: the specific way it fell over. The workaround someone built under deadline that somehow held for two years. How they got Finance and HR to agree on a single number without anyone updating their resume.

That knowledge doesn’t live in documentation. It lives in the people who already made the mistake you’re about to make. Rising is the one week a year they’re all in the same zip code. That’s the whole value. Proximity to people who’ve been burned the way you’re about to get burned.

Your permission slip.

So here it is, official: you are allowed to skip a session.

You are allowed to walk out of a breakout to keep talking to someone who’s three years ahead of you on the exact thing eating your Q4. That’s the highest-ROI thing you’ll do all week. In fact, buy them a Starbucks. You'll make a friend to boot.

Nobody’s manager has ever asked for the list of sessions they attended. They asked what you learned. Those are different questions, and they have very different answers.

The good ones have a little structure.

Not all conversations are equal. The random hallway bump is good. The organized one is better.

A roundtable. A small room. A group of people who do your exact job, one real question on the table, and nobody in the corner trying to sell you a platform. That’s the format that turns “nice to meet you” into “here’s the thing that just saved you six months.”

Seek those out. Some are on the official agenda (braindates, meetups). The best ones usually aren’t. (Keep your eyes open at Rising this year. There may be a room or two worth finding.)

Back to my ice pack.

I’ll be at Rising. Moving slowly, but I’ll be there.

Bad hip, good schedule.

I’ll be the guy you can actually catch, on account of I physically cannot run away from you.

Come have the conversation. That’s the whole product.

— Mike

Director HR Tech | Bad Hip, Good Schedule

P.S. When one of Justin’s prints fails (I have gallons of ruined plastic in my house), he doesn’t rewatch the official tutorial. He goes and finds the one person online who already broke it the exact way he just did. Ten minutes with someone who’s made your mistake beats an hour with the manual. He’s twelve and he already gets it. The manual tells you how it’s supposed to work. The person tells you how it actually breaks.

The Department of First Things First. For the people who do the work.

tldr;

You’re one of fifteen, twenty thousand people at Rising, not the person on stage. Stop trying to drink the firehose. You’ll drown and learn nothing.

The only artifact that justifies the trip is the one-page memo your boss actually reads. Reverse-engineer the entire week backward from that page.

Pick fewer sessions. Work the floor on purpose. Capture as you go. And for your own sake, pace yourself.

Somebody in your leadership chain approved the registration. Flights, hotel on the Strip, the badge. Four, maybe five thousand dollars, and a vague mandate that sounded like “go see what’s out there and bring back something useful.”

You are not keynoting. You’re not on a panel. You’re attendee number 8,742, and nobody rearranged the agenda around your calendar.

Good. That’s the job I actually know how to do. Here’s how you work Rising when the spotlight is pointed at somebody else.

Start with the memo. Work backward.

The mistake is treating the conference as the thing to consume. It isn’t. The thing you produce is the point, and that thing is one page.

Not a trip report. Not forty screenshots of slides you’ll never reopen. One page that says: here’s what I saw, here’s what it means for us, here’s what we should do about it in the next two quarters.

Write the headers of that page before you get on the plane. “What’s actually shipping.” “What we should pilot.” “What we should ignore for now.” And my favorite, “What did I learn that can provide immediate value.” (Try to find one). Every session you pick, every booth you stop at, is you filling in those four sections. If it doesn’t feed the page, it’s tourism.

Pick fewer sessions than you think.

The catalog is designed to make you feel like missing anything is a failure. It isn’t. Half of what’s announced from the main stage will be in the release notes by November, written more clearly and with none of the theatrics.

Skip one and use it to collect your thoughts. Instead, pick a session where a real customer stands up and says “here’s how this broke in our tenant and what we did about it.” Those don’t make it into the release notes. That’s the stuff you can’t get anywhere else.

Rule of thumb: 3-5 sessions a day, chosen well, beats eight sessions chosen out of FOMO. You are not there to fill a schedule. You’re there to fill a page.

The vendor floor is not your friend. Work it anyway.

Every booth is a funnel and you are the thing being funneled. That’s fine. You can still get value out of it if you walk up with a question instead of a lanyard-shaped hole in your soul.

Ask something specific. “Show me how your thing handles [the exact problem you’re fighting in your tenant].” Watch what they do when the happy-path demo runs out of road. That reaction tells you more than the booth graphics ever will.

And when they ask to scan your badge, decide before you’re standing there whether this is a real evaluation or a swag transaction. Both are allowed. Just know which one you’re in.

Capture in the moment, or you’ll lose it.

You think you’ll remember the good line from the breakout. You won’t. By Wednesday it all blurs into a beige hum of “AI-powered”, “agentic”, and free tote bags.

Whatever your system is (notebook, phone, voice memo in the hallway), the rule is the same: capture before the next session starts, not at the end of the day. The half-life of a conference insight is about ninety minutes. After that you’ve got a slide photo and a vibe.

Pace yourself. This is the one everybody blows.

Rising is four days. Some of us crazies come early and make it longer. You cannot do all of it on Monday. The people sprinting the vendor floor at 8am with three tote bags and a cold brew are the same people face-down in their hotel room by Tuesday afternoon, having peaked too early and absorbed nothing.

Skip a session block. Eat a real meal. Take your team to lunch and bond. Have the hallway conversation instead of racing to the next room. The thing you’ll actually remember in December is rarely the session you rushed to. It’s the twenty minutes you spent talking to someone who does your exact job at a company three times your size.

The page, one more time.

Four things. Not thirty. Your boss does not want the firehose either. They want to know what you saw, what it means, and what you’d do. Deliver that on one page while everyone else is still sorting through their camera roll, and you’re not attendee 8,742 anymore.

You’re the person who came back with a plan.

— Mike

Director HR Tech | Skips one Keynote on Purpose

P.S. This summer in Myrtle Beach, Justin found a store called Sugar Life and bought four pounds of candy in one shot. (I’m not innocent here. 1.3 pounds of gumballs. We all have our vice.) Twenty-four hours later: stomachache, mild regret, a working understanding of consequences. I keep telling Rising newbies the same thing, and I’ll tell you: it’s a marathon, not a sprint. Justin forgot that in Myrtle Beach. Don’t forget it in Vegas.

The Department of First Things First. For the people who do the work.

tldr; A portable, user-owned record of verified skills has been built at least four times, including by Workday. It keeps dying, and never for technical reasons. Three things kill it every time: the moat, no trusted signer, and cold start. The only place verification actually took hold is the one place the law removed the choice.

October 2019. Workday Rising. Workday announces the future of hiring: Workday Credentials, and a companion app called WayTo. Individually owned, verifiable credentials. Proof of what you can do, carried from company to company, shared on your terms. Bhusri pitched it on CNBC. They’d acquired an identity startup to build it.

I know, because I downloaded it.

It sat on my phone for about a year. I put things in it. And to be exact about what “things” means, because it matters: one credential, issued by Workday, certifying that I had attended Workday Rising. A digital badge proving I’d been in the room where the wallet was announced.

I tried to add my PMP. The app wasn’t ready for it yet.

That’s the graveyard in two sentences. A wallet whose only issuer was the company that built the wallet, and no path to load the credential I’d actually earned from an outside body whose entire business is attesting that people can do things.

Then one day it was gone. Not deprecated with a migration path. Not sunset with a notice. Just gone.

And no, this wasn’t a conference demo that evaporated the following Monday. Workday published Terms of Use for WayTo with an effective date in May 2021. Live legal terms, a real product, a design team, a user. I know when it launched, because I was watching. I don’t know when it ended, because nobody said.

It never got a goodbye.

If you have never heard of WayTo, congratulations. You just heard the whole story.

Your best idea has already been built

Every practitioner who stares at the arms race long enough lands on the same design: a portable, user-owned portfolio of verified skills, handed to an employer like a business card. I landed on it years after I’d already been issued one. So did everyone else.

MIT was issuing diplomas on Blockcerts in 2017. The W3C published an open standard for verifiable credentials. Velocity Network, a consortium stacked with names you’d recognize from every expo floor, launched a working mainnet in 2022.

The technology was never the bottleneck. Say that out loud the next time someone tells you we just need better tools.

“It didn’t die, it evolved”

Somebody is already typing that Credentials didn’t fail, it became Skills Cloud. Fair, and it’s the most important distinction in this series.

Skills Cloud infers. It reads your worker data and your job architecture and produces probabilistic skill tags. Genuinely useful for cataloging, internal mobility, workforce planning. I’m not knocking it.

But inferring is not verifying. A probabilistic tag is a guess with good manners.

And here’s why that matters. A candidate uses an LLM to generate keywords. The platform uses ML to extract probabilistic tags. A screening tool matches one probability against the other. That’s algorithmic noise matching algorithmic noise, and everybody gets a confidence score for their trouble.

Inference solved cataloging for the enterprise. It did not solve trust between a candidate and an employer. Different problems. Only one of them is drowning us.

Killer one: the moat

My favorite piece of evidence in this whole saga comes from LinkedIn’s own help documentation.

LinkedIn has a verification badge now. Real infrastructure, built on Entra rails. It verifies three things: that you are who you say, that you work where you say, that you went to school where you say. The docs are explicit about the boundary. Not your title. Not your expertise. Not one claim about what you can do.

They built the rails all the way to the edge of the thing that matters. And stopped.

Now: Follow the money. Talent Solutions, meaning seats and postings and pipeline tools, is the majority of LinkedIn’s revenue, and the moat under it is self-reported profile data plus network effects. Verified portable competence means employers search less, InMail less, buy fewer seats.

The ambiguity is the product. Verifying skills wouldn’t improve it. It would end it.

Indeed runs a different flavor of the same incentive: their model monetizes application volume. The flood drowning your recruiters is somebody’s revenue line.

Neither company is evil. Both are rational. If you read Part 1, you know rational is the problem.

Killer two: nobody trusted will sign

The moat only explains half the graveyard.

Workday had no moat conflict. Their customer is the employer. Portable verified credentials would have made the platform more valuable, not less. Clean incentives, and it still died. So did Velocity, a consortium built specifically so no single vendor could strangle it.

Different killer. A credential is only as strong as whoever signs it, and nobody trusted will sign the claim that matters. Ask a past employer to vouch for how good someone actually was. You’ll get dates and a title, in the flat voice of someone reading a script legal wrote. Because legal wrote it.

Certifications work fine for the certifiable. My (now renewing, but not all of them, that’s another story for another time) Workday Pro certs are verifiable right now, no blockchain required. But look at what they certify: knowledge of a system. The things that actually drive a hire (judgment, architecture instinct, the ability to tell a VP their timeline is fiction without getting escorted out) are the least certifiable things a person has.

The chain can carry any claim securely. The problem is upstream. Nobody credible will make the claim.

Killer three: nobody shows up first

Candidates won’t maintain wallets no employer checks. Employers won’t check wallets no candidate carries. Cryptography has never once solved a chicken-and-egg problem, and it didn’t here.

I was the guy who showed up first. Nobody was on the other side of the room.

The one place it actually works

Velocity isn’t dead, by the way. It found traction.

Guess where?

Healthcare. The one corner of the labor market where verification is mandated by law and every day of delay is a compliance problem and an empty shift. Where regulation removed the choice, the infrastructure works fine. Instantly, even.

So, the technology works everywhere. It gets adopted exactly where the law forces it. That’s an incentives finding, and it’s the whole thesis in one data point.

The defection that pays

So what breaks an equilibrium this stable? Regulation, which is healthcare’s story. A coalition big enough to move together, which is what Velocity tried and the cold start ate. Or defection that pays off individually, which is the hopeful one.

Justin runs a 3D printing business out of our basement. When a customer asks whether he can print something, he doesn’t send a document describing his printing philosophy and core competencies. He shows them the last thing he printed. Twelve years old, and he arrived at proof-of-work hiring on his own, because nobody has taught him the vocabulary that lets adults avoid it.

The research has agreed with him for decades. Work samples and structured interviews sit far above resume screening as predictors of job performance. Resume screening doesn’t crack the top of the list. The instrument we’re all pointing AI at is the least predictive one in the drawer.

One caveat, or I’m just moving the arms race down the funnel: the take-home is already compromised. Candidates finish your case study with the same tools they used on the resume, and they should, because you would. An asynchronous work sample in 2026 is a resume with extra steps.

So when I say work samples, I mean observed. Live, structured. Watching someone reason is the thing AI can’t hand them. The friction is the instrument.

And no, you can’t put 2,000 hourly applicants through a live evaluation. Nobody’s asking you to. The move is directional: shift weight off top-of-funnel resume filtering toward short structured evaluations that actually predict something. Nobody converts a whole pipeline by Q4, and anyone promising that is selling. Weight the proxy a little less each quarter and watch what happens to quality of hire.

Seven years ago the future of hiring was announced on a stage, and it quietly evaporated off my phone while almost nobody noticed. That’s a market telling you, in its own language, what it’s actually optimizing for.

Nobody profitable wants to fix it. Nobody willing to fix it can.

Until one of those sentences changes, the arms race is the roadmap.

-Mike.

Director HR Technology | Undertaker

P.S. — Going to Rising this fall? Ask a longtime Workmate, warmly, whatever happened to WayTo. Then watch their face. I don’t need to ask. I still have the muscle memory of opening an app that isn’t there.

P.P.S. — My friend Doug Barkholz died on August 28th. He was 54.

In my twenties, Doug had every single thing I was fairly sure I was never going to get. Twin little kids. A house that looked enormous to a guy whose furniture was mostly borrowed. Bev, who was the reason the whole operation ran. And he was, irritatingly, the best athlete in whatever room he walked into. Volleyball, tennis, ping-pong, pick a sport, he was better at it than you by the second set.

The way we all ended up in each other's lives is the part I keep telling people this week.

Jenn set the ball. Doug hit it. It hit Bev.

Broke her arm.

That is the actual origin story of one of the most important friendships of my adult life. A good set, a better swing, and an ER visit.

I watched Nathan and Samantha a lot of nights while the three of them played. I didn't think of it as babysitting. I thought of it as the price of admission to a house where people liked each other out loud, which was not something I had a ton of reference points for at the time.

Doug was born Tuan Quang Nguyen, in Vietnam. He got here the way I got where I am, which is to say somebody chose him. Adopted people find each other. Nobody announces it. You just end up in the same kitchen at 3am and something in the wiring recognizes something in the wiring, and five years later you're the guy watching his kids.

His obituary lists his mother and his birth mother in the same sentence, right next to each other. He'd have wanted it that way. Both of them are real. That's the whole thing.

I idolized him. I'm not going to soften that with something more dignified. I was twenty-something and I looked at his life the way you look at a floor plan.

What I understand now, and didn't then, is that he wasn't showing me a life to envy. He was making room in his.

Bev, Nathan, Sammy — you have the whole thing. All of it.

The crew has got you, Doug. Somebody else can set for Bev.

tldr;

Rising is six weeks out, and everything on that expo floor is designed to sit on top of your tenant.

None of the demos were built on your data. The audit is how you know your ground truth before somebody in a branded polo describes it to you.

Five checks, about thirty minutes each: job data, security model, integration inventory, business process reality, and the why coverage.

Each check arms you with one question that separates a real product from a happy-path demo.

Half a day now. It’ll be the highest-ROI half day of your conference. I’ll be there; come argue with me.

Six weeks to Rising.

The session catalog is live, your inbox is filling with “will you be in Vegas?” emails from vendors whose products you cannot name, and somewhere a demo team is rehearsing on a tenant so clean it should be in a museum.

Here’s the thing about everything you’re going to see in that expo hall: all of it, every agent, every experience layer, every AI-powered anything, is designed to sit on top of your tenant. And none of it, not one demo, was built on your data.

The gap between those two facts is where implementation hangovers come from.

So before you fly, run the audit. Five checks. Thirty minutes each. The goal is not to fix anything (you won’t, not in six weeks, and that’s fine). The goal is to know your ground truth before someone else describes it to you, because the practitioner who knows their ground truth asks a different kind of question at the booth. The kind demo teams practice avoiding.

Check 1: Pull ten job profiles at random

Not your ten best. Ten random ones. Read them like a candidate would, or like an AI agent will.

When were they last touched? Do they describe the job as it exists, or the job as it existed when someone copied it forward during implementation? Do similar roles use similar language, or did five recruiters invent five dialects?

Why this one’s first: every AI capability you’ll see demoed this year (matching, sourcing, talent summaries, skills inference) runs on this data. The demo runs on curated profiles. Yours will run on whatever you just pulled.

THE BOOTH QUESTION: “Show me what your product does with a job profile that hasn’t been updated in four years. Not a cleaned one. That one is my tenant.”

Check 2: Pick one manager. Enumerate what they can see.

One mid-level manager. Trace their actual access: security groups, role assignments, the reports they can run, the data those reports expose. Not what the security matrix says on paper. What the tenant actually resolves.

If that exercise takes you more than thirty minutes, that’s a finding. If the answer surprises you, that’s a bigger finding.

Why it matters this year specifically: the biggest pitches at Rising are experience layers and universal search. One conversational front door across all your systems. Every one of those pitches is a stress test of your security model, because search surfaces whatever access allows, at scale, without the natural obscurity of nobody-knows-where-to-click.

THE BOOTH QUESTION: “Walk me through what happens when your search layer meets a security group that was misconfigured in 2021 and nobody noticed because the data was three menus deep.”

Check 3: Produce the integration number

How many integrations are running against your tenant right now? Not roughly. The number.

If you can’t produce it, you’re in good company (I have walked into organizations that genuinely could not give me a number), but you can’t orchestrate what you can’t inventory, and you definitely can’t budget for it. While you’re counting, mark which delivered integrations have been modified, because under the current Flex Credits policy language, modification changes how they’re metered. That’s not commentary, that’s just what the contract says now, and every new product you evaluate at Rising drinks from the same wallet.

THE BOOTH QUESTION: “What does your integration actually consume, in whose credits, and what happens to that math when we customize it? Show me the meter, not the diagram.”

Check 4: Count the steps in your three highest-volume BPs

Pick the three business processes that run the most: probably hire, job change, and one of your absence or comp flows. Count the steps. Then mark the ones that exist because of a workaround somebody built in 2019 and everyone has been stepping around ever since.

Why: the agent pitch this year is process automation. An agent can’t optimize a process nobody can accurately describe, and it absolutely will automate your workaround at machine speed if you hand it one. Garbage in, garbage out, but faster.

THE BOOTH QUESTION: “Before your agent optimizes this process, how does it find out that step six only exists because our old payroll vendor couldn’t handle mid-period transfers? Because nobody wrote that down.”

Check 5: Run the why coverage

Last one, and if you’ve been reading this newsletter all summer you saw it coming. Take your most load-bearing configuration (the calc field everything depends on, the condition rule with the weird exclusion, the modified integration from check 3) and ask the Bus Factor question: who else knows why this is the way it is?

If the answer is nobody, write the ledger entry now, from memory, dated and flagged as reconstructed. Four lines. Because here’s what happens after Rising: you come home with proposals. Change lands hardest on tenants where the why is undocumented, and you’re about to propose change.

THE BOOTH QUESTION: This one’s not for the vendor. It’s for you, on the flight home, reading your own notes: “does this proposal survive contact with the tenant I actually have, or the tenant I described in the meeting?”

Half a day

Five checks. Call it half a day, generously. What you get back: a ground truth nobody at Rising can hand-wave past, booth questions that end happy-path demos early, and (the quiet benefit) the beginnings of the case for whatever you actually do want to buy, because “our job data is clean, our security model is verified, and here’s our integration inventory” is how funding conversations start.

Clean your room before you shop for robots. Same religion, conference edition.

And yes: I’ll be at Rising. A couple of rooms, a lot of hallways. Details on exactly where in a few weeks, but if you run this audit and want to argue about what you found, come find me. That’s my favorite genre of conversation.

— Mike

Director HR Tech | Ground Truth Department

P.S. — I found out during the first round of the mixed doubles at Arthur Ashe. Between matches, the DJ played "Jolene." Twenty thousand people who'd come to watch tennis sang and danced to every word, and nobody had to ask them to. That's as good an epitaph as I've ever heard a stadium PA deliver. She gave away more than she ever took, stayed kind for six decades whether anyone was watching or not, and never once made a fan feel small for being one. Rest now Dolly, and thank you.

The Department of First Things First. For the people who do the work.

tldr; The resume was never a measure of competence. It was a proxy that survived because faking it took effort, and AI removed the effort from both sides at once. We now screen with AI and act surprised when candidates apply with AI. That is not an ethics problem, it is a prisoner’s dilemma, and you cannot ethics your way out of one.

“We’re drowning in AI-generated resumes. Every application looks perfect and they all read identical.”

Ten minutes later. Same meeting. “Okay, next item. The demo of the new AI screening tool.”

Nobody laughed. Nobody was supposed to.

That meeting is happening everywhere

I want to be careful here, because it would be easy to write this as a story about dumb decisions, and it isn’t one. Everyone in that meeting is behaving rationally. Hold that thought, because it’s the whole point.

First, the numbers, because this stopped being a vibe a while ago.

LinkedIn is processing roughly 11,000 job applications per minute, up 45% year over year, per reporting in the New York Times. Recruiters describe a “1,000 applicant threshold” where a single competitive req clears a thousand applications in days. Around 90% of HR managers say their workload has gone up because of the AI application surge.

And here’s the number that should end every “AI is making hiring more efficient” sales call: per SHRM’s benchmarking, time-to-hire and cost-per-hire have both risen for three straight years. Right alongside the rollout of generative AI tooling on both sides of the funnel.

Read that again. Everyone is spending more money to hire more slowly. That is the arms race’s actual output. Not a side effect. The output.

The resume was already dead. AI just called the time.

Here’s the thing: the resume was never a measure of competence. It was a proxy. A lossy, self-reported, unverifiable proxy that survived for one reason only. Faking it took effort.

That friction was the entire security model. Not verification. Not validation. Just the assumption that a person wouldn’t spend forty hours tailoring lies for one application.

AI removed the friction. In both directions, at the same time. Candidates can now generate a keyword-perfect resume for any req in the time it took you to read this paragraph. Employers can generate a plausible-sounding rejection for ten thousand of them before lunch.

So both sides are now optimizing a document that neither side believes.

You cannot out-detect a dead signal. There is no screening tool sophisticated enough to extract truth from an artifact that no longer contains any. The vendors selling you AI to “cut through the noise” are selling noise-cancellation for a room where the signal already left.

We did this to ourselves, and I can prove it with one sentence

Here’s the sentence: We screen candidates with AI, and we are shocked, shocked, when candidates apply with AI.

That’s not an ethics problem. Run the symmetry test on any rule in your hiring process. If a rule doesn’t survive being reversed, it was never a principle. It was just leverage. “AI is fine when we use it to reject you, but cheating when you use it to reach us” fails that test so hard it should come with a laugh track.

And once both sides are armed, the escalation logic takes over. This is a textbook prisoner’s dilemma. Every individual move is rational. The TA leader can’t stop posting where the candidates are. She can’t hand-review a thousand applications with four recruiters. The candidate can’t unilaterally disarm either, because he knows a bot reads his resume before any human does. Each player’s best individual move sustains the worst collective outcome.

And the defining feature of a prisoner’s dilemma is this: you cannot ethics your way out of it. Responsible-use pledges don’t beat incentives. They never have. Which is an awkward thing to say out loud in a year when “Responsible AI” is on every conference agenda. I’ll say it anyway.

The part that actually turns me green

It gets worse, because employers aren’t just trapped in this equilibrium. We fund it.

Every recruiter seat license. Every ATS module. Every screening add-on, every sourcing tool, every “AI-powered talent intelligence platform.” All of it is revenue flowing to an intermediary economy whose business model depends on the resume staying exactly as ambiguous and unverifiable as it is today.

We are paying protection money to the arsonist. And buying our smoke detectors from them too.

(Who profits from the ambiguity, who quietly built the fix and let it die, and what LinkedIn’s own verification badge accidentally confesses: that’s Part 2. Bring a shovel. We’re touring a graveyard.)

The concession, because this newsletter isn’t a whine

If you run TA and you just signed a PO for a screening tool, I’m not calling you a sucker. The volume is real. The four-recruiters-and-a-thousand-apps math is real. The tool genuinely stops your team from drowning this quarter, and this quarter is when your requisitions are due.

That’s what makes it a trap. Traps built out of stupid decisions are easy to escape. Traps built out of rational ones are a different deal entirely.

And one more concession, ahead of Part 2: the research on what actually predicts job performance has been sitting in plain sight for decades. Observed work samples. Structured interviews. The Schmidt-Hunter line of research puts work sample validity miles ahead of resume screening, which doesn’t even crack the top predictors. (Not take-home assignments, before anyone writes in. Those fell to the same tools the resume did. Part 2 deals with that.) The better instrument exists. It’s just slower, and nobody got fired for buying the fast one.

So no, the arms race isn’t a failure of technology. The technology is working great. It’s a failure we’ve collectively decided to subscribe to. Annually. With auto-renew.

Back to that meeting from the top. The person who described the flood and the person who approved the flood-management tool were the same person. That’s not irony.

That’s the system. Working as designed.

— Mike.

Director HR Technology | Battle-Bot referee

Coming soon: the graveyard tour. Everyone who built the fix, including a name you know very well, and why the fix keeps dying.

P.S. Yes, I used AI to help research a piece about AI ruining hiring. The difference is I’m telling you.

tl;dr

• SHRM still sells hope to students (”backpack to boardroom,” $49 a year) and sells fear to the professionals who already paid in (the June SHRM26 keynote: HR faces “extinction”).

• The hope is the acquisition layer. The fear is the retention layer. Which one you get depends on whether your dues have cleared.

• In December a Colorado jury hit SHRM with $11.5M for racial discrimination and retaliation, and the judge let them weigh SHRM against the very expertise it sells. The court refused a retrial this spring. SHRM is appealing.

• I had a concession ready in June: tell the kid to join anyway, the scholarships are real. I’ve spent two months unable to write that sentence.

Years ago, at Temple’s Fox School of Business, SHRM was the cool kids club.

I’m not being cute about it. If you were in the business school and you so much as looked sideways at HR, SHRM was where the serious people were. The chapter had a vibe. The badge meant something. It meant you weren’t just taking a class, you were joining a profession. It meant arrival.

I wanted in. So I got in.

Then I graduated, got busy, got broke, and let it lapse. I drifted first. This was never going to be a story about SHRM leaving me.

Here’s what dragged me back to it. In June, at SHRM26 in Orlando, Johnny Taylor took the main stage in front of video walls full of humanoid robots and told twenty thousand HR professionals that their profession had “lost the plot” and was staring down extinction. The fix, per the keynote, was to rebrand yourself a “Chief Work Officer” and start conducting the robots before they conduct you.

And it occurred to me that I had no idea what SHRM even sells anymore. So I went and looked.

It sells two completely different things. Which one you get depends entirely on whether you’ve paid.

To the professionals in that ballroom, it sells dread. And not for the first time. 2024 was “a storm is coming for HR.” 2025 was “another monster year,” the storm now officially here. 2026 is extinction. Three years of the same escalating weather report from the same stage. If your product is HR’s anxiety, business is booming.

To the kid I used to be, it still sells hope. Aggressively. More than when I was at Fox. There’s a slogan now, “backpack to boardroom.” Student membership runs forty-nine bucks, pitched as an investment in your future. There’s a discounted two-year runway to walk you from student into full paying dues, and schools can bulk-buy memberships ten students at a time. Four hundred grand in scholarships. A mentoring cohort.

Sit with that split for a second. The eighteen-year-old gets arrival. The forty-eight-year-old gets survival. Same organization, same summer, opposite emotions, sorted by tenure.

That’s the actual business. Hope is how they get you in the door. Fear is how they keep you paying once you’re inside. The cool kids club I remember didn’t disappear. It moved to the front of the funnel, where the new customers are.

I could have published a version of that in June. I nearly did.

Then I sat with the other thing.

In December, a Colorado jury found SHRM liable for racial discrimination and retaliation against a former employee named Rehab Mohamed, and set the number at eleven and a half million dollars. Ten of that was punitive, which is a jury’s way of saying this was not a close call. This spring the court refused to give SHRM a do-over. They’re appealing.

Read the details and it gets worse. The internal investigation was run by someone who, by his own testimony, had never conducted a workplace discrimination investigation before. One training session. The organization that sells the certification, the toolkit, the webinar, and the breakout session on how to handle exactly this could not handle exactly this.

And the judge let the jury weigh SHRM against its own reputation. The court ruled SHRM’s standing as the authority on HR could not “reasonably be excluded” from the case. So a jury got to ask the question out loud: does the body that grades everyone else’s homework pass its own? It said no, eleven and a half million times.

Here’s the part that reframed the keynote for me. That verdict landed in December. The retrial was denied in the spring. So when Taylor stood in Orlando in June and told a room full of practitioners they had lost the plot, the house he was speaking from had already been told, by a federal jury, that it had lost it worse.

Now the concession, because a piece like this owes you one.

The scholarships are real money. The mentoring is real. The local chapters and the volunteers who run them are some of the best people in this field, and they had nothing to do with any of the above. A verdict against the national org is not a verdict against the person running your regional chapter for free on a Tuesday night. Keep those straight.

But here’s where I get stuck.

Two months ago I had a paragraph sitting right here that told the twenty-year-old to join anyway. The scholarships are real, I wrote. The network is real. Go get it.

I’ve tried to keep that paragraph. I can’t.

You don’t learn fairness from the house that just paid eleven and a half million dollars for forgetting it. The backpack deserves better than what’s waiting in the boardroom. Go look at other institutions like HRCI.

- Mike.

Director HR Technology | Highly disappointed

P.S. - Help me out, what other places should we send students to learn about the human resources profession?

tldr;

• Two people I respect and myself have three different names for the same job. I assumed we were all saying the same thing. We are not.

• Unicorn is a scarcity claim. Builder is a craft claim. HRIS gang is a solidarity claim. Each one is a different theory of why the work goes unseen.

• The real tell isn't which nickname you use. It's that the nicknames exist at all.

• Settled professions don't need affectionate in-group slang for themselves. Ours grew in the gap our official titles left behind.

• A field that has never had to defend itself as a discipline doesn't get to name itself. That's a problem we're about to feel in a much more expensive way.

Ceci calls us unicorns.

I call us builders.

Dani calls us the HRIS gang.

Three people, same discipline, three different words. I noticed it and thought, well, that's nice. Everyone has their own affectionate shorthand for the same thing. We're all basically saying the same thing.

We're not.

Three names, three arguments

Unicorn is a scarcity claim. You are rare. The market cannot reliably produce you. There is no degree program that outputs a person who can hold configuration, compliance, org design, and a straight face in the same meeting. If you're a unicorn, your value comes from how hard you are to replace. The compliment carries a warning inside it: rare things are also lonely things, and there is no bench behind you.

Builder is a craft claim. That's mine, and I'll own the bias. It says the identity lives in the making. You didn't advise on the thing. You didn't recommend the thing. You built the thing, and then you owned it at 11pm when it broke. Craft claims put the dignity in the work itself rather than in the title above it. That's the appeal. That's also the limitation, because craft claims don't scale into org charts very well.

HRIS gang is a solidarity claim. It's us. And any time there's an us, there's a them, and the them is usually the layer of the org chart that describes your job inaccurately in front of people who can fund it. Solidarity claims are the warmest of the three and the most honest about the emotional reality of the role. They also, quietly, concede something. You don't form a gang when you're already at the table.

Rare. Productive. Embattled.

Pick one and you've picked an entire theory of why this job feels invisible.

The part I actually can't stop thinking about

Here's what got me, though, and it isn't which name is right.

Professions with a settled identity do not need affectionate slang for themselves.

Surgeons are not the scalpel gang (although that is a far cooler name). Structural engineers do not refer to each other as unicorns, and they are genuinely rare and genuinely load-bearing. Actuaries are as niche as it gets and have never once needed a nickname to feel real. Those fields have a name, a body of knowledge, a credential that means something specific, and a shared story about what the work is for.

We have five nicknames and counting.

The nicknames didn't appear because we're clever. They appeared because the official language never described the work. "HRIS Analyst" describes maybe a fifth of what the job is now. "HR Technology" is a category, not a practice. "Systems Administrator" is a 2009 title on a 2026 job. So people built their own vocabulary in the gap, and the vocabulary is affectionate because it had to do emotional work that the titles never did.

The slang is a symptom. It's what grows in a discipline that never got formally defined.

Where the concession goes

Now, I don't want to be precious about this, because the nicknames are load-bearing and they've earned their keep.

"HRIS gang" got a lot of people through month 14, when the SI statement of work ended and the phone stopped getting answered. That kind of belonging is not decoration. It's retention. I've watched people stay in this field because of a group chat.

And there's a fair defense of the naming mess itself. The work genuinely changes every eighteen months. Position management, integrations, security design, and now agent governance. A field moving that fast has a real reason not to have settled on a name yet. That's not entirely a failure. Some of it is just velocity.

Fine. Both things are true.

Why this matters more than it used to

But here's where it stops being a fun vocabulary observation.

I've been writing about how competence gets proven in this industry. Screening systems reading resumes that other systems wrote. Credentials that live on a platform until the platform decides otherwise. The whole apparatus of demonstrating that you can actually do a thing.

You cannot credential a discipline that has not defined itself.

You cannot build a portable, verifiable record of skill in a field where five practitioners give you five different accounts of what the work even is. Every certification we have certifies product knowledge, which is exactly the thing that expires. None of them certify the actual practice, because the practice has no agreed shape. The nickname problem and the credential problem are the same problem wearing different clothes.

That's the bill coming due. Not today. But soon, and from a direction most of us aren't watching.

So

I'm not going to tell you to stop saying unicorn. I like unicorn. Ceci is right that we're rare and she's right that the market can't manufacture us.

And I'm not going to tell you to stop saying HRIS gang for similar reasons. There is and us and them right now, and we need the group emotional support.

I'm saying the reason we have different names is that we've never had to defend this as a discipline. Nobody made us write down what the work is, what good looks like, or what someone should be able to do before they're trusted with a tenant. We got to skip that step.

And a field that has never had to defend itself doesn't get to name itself.

So pick your word. Unicorn, builder, gang, whatever you've got.

Then be ready to say why.

— Mike.

P.S. I want to hear yours. What do you actually call this job when you're talking to someone who does it too, not the title on your badge? Reply and tell me. I'm collecting them, and I'm going back to Ceci and Dani to ask why they picked theirs, because I think the reasons are more interesting than the words.

tldr;

A scenario: an AI recommends a comp range bump, everyone nods, and six months later there’s an equity issue nobody can trace. It’s hypothetical. The gap it exposes is not.

The dangerous tier of AI in your tenant isn’t agents taking action. It’s recommendations that humans rubber-stamp. That’s action with extra steps and the accountability boiled off.

Governance is four named answers before any recommendation stream goes live: who owns the objective, who reviews, how a human challenges it cheaply, and what the rollback is.

Plus one ledger entry every time an accepted recommendation changes a standing number. The system logs the transaction. Somebody has to log the decision.

When it goes wrong, it will not be an AI failure. It will be a governance failure. Those have owners.

Let me give you a Tuesday morning.

Your tenant is running an AI capability that analyzes talent pipeline data and suggests comp range adjustments for open requisitions. This is real functionality. This is the good stuff. This is what the roadmap slide promised.

The agent suggests moving a range from $85K to $105K. Market data, pipeline conversion rates, a tidy confidence figure sitting next to the number.

The recruiter sees it and thinks: cool, the system says so. Adjusts the range.

The hiring manager sees the adjusted range with a system recommendation attached and approves it, because approving things with system recommendations attached is what Tuesday is for.

The comp partner is cc’d and says nothing, because it came from the tool that Finance signed off on last year.

The offer goes out at the new range. The next req in that job family anchors on it. So does the one after that. Every subsequent approver inherits a number that arrived pre-blessed.

Nobody asked: what market data? Which pipeline, over what period? What was the model optimizing for? Time-to-fill, quality of hire, and retention are different targets, they conflict, and the model picked one. Somebody in your organization needs to know which one, and why, and nobody does.

Six months later you have a comp equity issue you cannot trace, because it started with a recommendation that twenty people approved by not questioning it.

The dangerous tier is not the one you think

Everyone’s governance anxiety points at the scary tier: autonomous agents taking action in the tenant. And fine, that tier deserves respect. But it also gets respect. When an agent can execute a business process end to end, people instinctively demand approvals, logging, kill switches. Fear does the governance work for you.

The recommendation tier gets none of that, because recommendations feel safe. A human is “in the loop.” Except here’s what the loop actually looks like: we are phenomenal at deferring to something that sounds confident. Put a percentage next to a number and watch a whole meeting nod like it came down a mountain on a stone tablet. The research on automation bias has been saying this for decades. Your own last vendor demo said it louder.

A recommendation that humans reliably rubber-stamp is an action. It’s an action with extra steps, and the extra steps exist mainly to distribute the accountability until it evaporates. When the agent acts, the agent’s owner answers for it. When twenty people each add a nod, nobody answers for anything. The loop isn’t a control. The loop is a laundering mechanism.

So the question at this tier was never “can the AI do it?” It’s “can the human evaluate it?” And evaluation is a skill your org has to deliberately build and deliberately protect, because the default trajectory of every recommendation stream is toward the Tuesday morning above.

Why you can’t trace it

Notice the specific shape of the damage: not that the number was wrong. The number might have been right! Maybe $105K was correct for that req, that week, that market. The damage is that six months later, nobody can reconstruct why it happened, whether the conditions that justified it still hold, or how far the anchor has drifted through the job family since.

The system logged the transaction. Every click, every approval, timestamped and audit-ready. And none of it helps, because the audit trail records who touched it, not why anyone believed it. If you read the Why Ledger post a few weeks back, you already know where this lands: the what is everywhere and the why is nowhere. An accepted AI recommendation that changes a standing number is exactly the kind of decision that earns a ledger entry. Four lines. What the model claimed, what the human checked, what would have changed the answer. The day the equity question arrives, that entry is the difference between an explanation and an excavation.

Who holds the leash

Here’s the governance model that fits on an index card. Before any AI recommendation stream goes live in your tenant, four questions get answered with names. Not roles. Names.

Who owns the objective? Somebody approved what this model optimizes for, and knows what it trades away. If the answer is “the vendor’s default,” that’s an answer, and it’s a bad one, and now it has an owner anyway.

Who reviews? Not every recommendation, that’s fantasy. A sampling cadence. Somebody pulls ten accepted recommendations a month and interrogates them like they haven’t been pre-blessed.

How does a human say no, cheaply? If challenging the system costs a recruiter an awkward escalation and a reputation for being difficult, nobody will ever challenge the system. The override path has to be cheaper than the nod, or the nod wins every time. This is the question almost every org skips.

What’s the rollback? The model was right in March and wrong in September because the business changed and nobody told the model. Who notices? What’s the drift review? How do you unwind an anchor that’s already propagated through a job family?

Four names, a sampling cadence, a cheap override, a rollback plan, and a ledger entry per accepted change. That’s not a transformation program. That’s a Tuesday afternoon of decisions. You could have it done before the next release drops.

While it’s still hypothetical

Has the comp scenario actually happened? Not in my tenant. Not that I know of, which, if you’ve been paying attention, is exactly the problem: the whole failure mode is designed to be invisible until somebody goes looking.

The agents are coming either way. The recommendations are mostly already here, sitting in your tenant with tidy confidence figures, being nodded at. The dog always sees a squirrel named efficiency, and it just goes. Nobody’s arguing about the dog anymore.

The only open question is the leash, and the best time to answer it is while the scenario is still a scenario. Because the worst time to build a governance model is the Monday morning after you needed one.

— Mike

Director HR Tech | Keeper of the Leash

P.S. Justin’s sand volleyball club has exactly one governance rule: when the ball hopper runs low, everybody goes out and shags balls. Last week, nobody did. Not defiance; everyone just figured someone else had it. Play stopped, and a dozen kids stood on the sand looking at each other over an empty hopper until Coach snapped them out of it. It looked like the Spider-Man meme for a second. A rule assigned to everybody is a rule assigned to nobody. Twelve-year-olds learn this in one afternoon. We build recommendation streams on it.

The Department of First Things First. For the people who do the work.

tldr; The Silver Lake news isn’t the change. It’s the confirmation.

• Reuters reported talks, not a deal. ~$43B valuation, no terms, real chance it never closes.

• Workday already executed the PE playbook itself: 1,750 cuts in Feb 2025, ~400 more this February weighted toward customer support, consumption pricing via Flex Credits, growth decelerating into a margin story.

• Silver Lake wouldn’t arrive to impose discipline. The discipline is already installed. That’s the part worth worrying about.

• What a close actually changes for you: roadmap statements lose their accountability, the community layer gets a return threshold it’s never faced, and anything unmonetizable inside the hold window quietly slides.

• Do three things now: treat roadmap slides as risk items, convert verbal commitments to paper while pre-close leverage is at its peak, and inventory what you own versus what you rent.

• The fair counterargument: the public market has been a bad owner, and architectural work is genuinely easier without a ninety-day clock.

Yesterday afternoon, WDAY got halted.

By the time it reopened it was up double digits (the best single day the stock has ever had) and my phone was doing the thing where three separate practitioner group chats go off at once. Reuters had the story: Silver Lake in talks to take Workday private at something around a $43 billion valuation. Talks. Not a deal. No signed anything.

Within about ninety minutes, LinkedIn had already written the eulogy.

You know the post. You’ve probably drafted a version of it in your head. PE buys software company. PE cuts R&D. PE guts support. PE raises your renewal nine percent and calls it a partnership. It’s a good post. It’s mostly historically defensible.

It is also, in this specific case, about eighteen months late.

Here’s the thing nobody reacting to Thursday wants to sit with. Workday has already been running the private equity playbook. On itself. In public. With a filing requirement.

Count the moves

February 2025: 1,750 roles cut.

February 2026: roughly 400 more, weighted toward customer support and roles described as non-revenue-generating.

Read that phrase again. Non-revenue-generating. That is not SaaS-company language. That is deck language. That is the vocabulary of somebody building a bridge to a margin target.

Flex Credits: consumption pricing bolted onto the AI surface area. Sana. Pipedream. A product story that reorganized itself around one monetizable narrative in about four quarters flat.

Subscription growth decelerating from the high teens toward fourteen percent, with the gap between “growth story” and “profitability story” closed the way that gap always gets closed.

None of that required a sponsor. That was management, reading the same room the sponsor reads.

So when people say Silver Lake is going to come in and cut: cut what, exactly? The support org already took the hit. The consumption model is already installed. The AI-or-nothing product thesis is already the entire roadmap. Silver Lake wouldn’t be arriving to impose discipline. They’d be arriving because the discipline is already there, the multiple compressed anyway, and that is a very attractive combination if you have $43 billion and a ten-year clock.

The scary version isn’t what changes. It’s what doesn’t have to.

That’s the reframe.

We’ve been trained to fear the moment of acquisition (the announcement, the integration memo, the new logo on the support portal). But in situations like this, the damage lands in the two years before the deal, when a public company starts quietly optimizing itself into something buyable.

We watched that happen. Most of us filed a ticket during it.

What a close would actually change is narrower, and worse in a quieter way.

Roadmap commitments stop being commitments. Not because anyone’s lying: Because the disclosure obligations that made a forward-looking statement semi-accountable disappear with the ticker. Everything your AE showed you in a sales cycle becomes, formally, a nice conversation you had.

The unmonetizable stuff gets a review it has never had. The community layer (RUGs, Customer Sharing, the parts of Rising that aren’t a sales motion, the advisory structures) has never been asked to justify itself against a return threshold, because it never had to. Under a sponsor with a hold period, everything gets asked. Community is the easiest line item in the world to defend philosophically and the hardest one to defend in a model.

Anything that can’t monetize inside the hold window dies quietly. Not killed. Deprioritized. Sequenced. Moved to future consideration. You’ll recognize the language when you see it, because you’ve seen it.

What you actually do about it on Monday.

Stop treating roadmap slides as planning inputs. If a project plan of yours has a dependency on functionality that is currently a screenshot from Rising, that dependency is a risk item now. Write it down as one, with an owner and a date.

Get what you were promised onto paper while the leverage is good. Right now Workday needs clean quarters, clean retention, and zero customer noise more than it has in a decade. That is exactly when a renewal conversation goes your way. If you have an ask (pricing protection, a Flex Credits floor, a written delivery commitment) the window is open, and it will not stay open through a close.

Inventory what you own versus what you rent. Every custom integration, every calculated field nobody documented, every Studio job with one person’s name on it. If the vendor relationship gets more transactional, internal capability is the only asset you hold that doesn’t reprice.

Where the doom take is wrong

The “PE ruins everything” version is lazy in one specific way, and I’d rather name it than pretend it isn’t there.

Silver Lake is not a strip-and-flip shop. Dell, EA, Endeavor: those are long-hold, operationally involved bets, not SolarWinds-style margin surgery. And there is a real argument that the public market has been an actively bad owner for Workday: punished it for investing in AI, punished it for decelerating, punished it for not being an AI-native startup that didn’t exist in 2005, and left the stock forty percent off its peak while management did roughly the right things.

Multi-year architectural work is genuinely easier without an earnings call every ninety days. That’s not spin, and if you’ve ever watched a quarter-end kill a good idea inside your own shop, you already know it’s true.

The question isn’t whether private ownership could fund the rebuild. It’s whether it would, or whether it goes where that freedom usually goes.

Based on the last eighteen months, place your bets.

And it might not happen at all

Worth saying plainly, because the internet skipped this part on the way to the eulogy: these are talks. Reuters said talks. A deal at this size, at this valuation, with this much financing to assemble, has a real chance of never happening. Half the takes you read this week will be about an event that never occurs.

But an eighteen percent pop on a rumor is its own data point.

The market just told you what it thinks Workday is worth the moment it stops having to explain itself.

That’s the news.

- Mike.

Director HR Technology | Not an Undertaker

P.S. - If this thing signs before Rising, that conference is going to be the strangest one we’ve had in years. Half the sessions built around a roadmap, half the hallway conversations quietly asking whether the roadmap survives. I’ll be there either way. Come find me, and bring the version of your opinion you wouldn’t post.

tldr; I counted every answer to “what do you call yourself.” The most common one wasn’t a title. It was an if-statement.

• Four in ten gave an audience-conditional answer: “if they’re in HR I say X, if not I say Y”

• Two in ten used a straight functional title, mostly the senior folks

• One in ten for fixer language, one in ten for payroll-outcome framing

• One in seven were pure bit

• A handful couldn’t answer at all, and meant it

• The router you built works great on humans. It doesn’t work on agents.

Two weeks ago I asked what you call yourself. I expected a laugh and maybe twenty comments.

I got 91, and then I got numbers, which is the part nobody warned me about. Because once you have data, you have to stop telling the funny version of the story and start telling the true one.

So I counted. Here’s what’s actually in there.

The finding I didn’t expect

The single biggest group (call it four in ten of you) didn’t give me a title at all. They gave me a conditional statement.

“If they’re in HR, I say X. If they’re not, I say Y.” Over and over, with the variables swapped: IT for HR, HR for IT, “computers” for family, “accounting” at parties. Amélie Marzin-Germain runs three different answers depending on who’s in the room. Sakshi Vaid Srivastava’s family still thinks she “fixes people’s computers.” Aleefa Shakeel gets it from both directions (IT according to HR, HR according to IT) in the same sentence.

So essentially, you’ve built a small router that takes “audience” as input and returns “acceptable answer” as output, and you run it so fast you don’t notice you’re doing it anymore.

Nobody self-reported this as a system. Everybody described it as an inconvenience.

The rest of the tally

Straight functional titles, stated with no hedge (HRIS Manager, HR Systems Partner, HRIS Systems Strategist) came in around two in ten. I’m guessing these are mostly the senior, established folks. Enough tenure that the confusion has calcified into a title and nobody argues with it anymore.

Fixer language (”the one who fixes it when it breaks,” Aldair Maldonado’s “mechanic you bring the car to”) was about one in ten. Smaller than I expected, given it’s the answer that gets the most reactions when someone says it. High comedy, lower incidence. People love reading it more than they use it.

Payroll-outcome framing (”I make sure you get paid,” Charlie Spitzer’s “I pay people and make sure they don’t die without benefits” pure gold, btw) another one in ten, roughly. This one’s interesting because it’s the only category that skips the org chart entirely and answers in terms of what the person on the other end actually gets. No mention of HR, IT, or Workday. Just the outcome.

Pure bit (no real answer, just the joke) landed around one in seven. Cliff Jurkiewicz went with “I am Spartacus.” (although it was less bit, and more commentary on our profession once you toured his “rabbit hole”). Shawn Latta and Adam Calvert both reached for Friends independently (transponster, Chandler Bing - funny how it all circles around the dude who had a job that no one can name). Sara Ayers gave two: “System Therapist” and “Urgent Chaos Coordinator.” Small, specific note: three different people, unprompted, reached for a therapist metaphor. Diane G., Sara Ayers, and Laura Horton’s “Tenant Therapist” all landed on the same idea from different directions. Make of that what you will. I have thoughts. They’re for a different issue.

And then there’s the smallest group, which is also the most important one. A handful of you didn’t answer the question at all. Paulo Poinha’s response was basically “what am I, actually” (asked in earnest, not as a bit). That’s someone hitting the actual wall the first issue was about, in real time, in a comment box.

What this does to the original argument

The first issue said: we name ourselves after tools or failure modes because we’ve never written down our scope.

The tally sharpens that. The biggest group of you didn’t even get as far as picking a wrong name: you built an audience-detection system instead, and you did it so quietly you don’t count it as an answer. That’s a more sophisticated coping mechanism than “I’m a Workday admin,” and it’s also a more expensive one, because a router has no scope by design. It’s not supposed to have one. It’s built to say whatever gets a nod back.

Here’s the fair complication, and it matters: for a lot of you, the router isn’t a failure: it’s a translation skill you’ve built on purpose, and it’s genuinely useful when you’re the only person in the room who can talk to both HR and IT without an interpreter. I’m not telling you to stop doing it. Codeswitching between departments is a real skill, not a symptom.

The problem isn’t the router. It’s using the router as a substitute for ever writing the scope down anywhere. You can keep three audience-specific answers for cocktail parties and still have one written answer for the org chart. Most of you have the first. Almost none of you have the second.

The one answer the router can’t return

Here’s what bothers me, now that I’ve sat with these numbers for two weeks.

The router works. That’s the problem. It works well enough that nobody in this function has ever been forced to answer the question once, in writing, in a form that holds for every audience.

So let me put a fixed return value on the table, and you can tell me I’m wrong in the comments.

Finance doesn’t run a router. Finance owns the ledger. There’s a book, the book is authoritative, and every downstream number in the company resolves back to it. Nobody asks Finance what they do, because everybody already knows what Finance holds.

We hold one too. Not “HR data.” Not “the tenant.” The authoritative definition of how work is structured: what a job is, what a position is, who reports to whom, when someone becomes a worker and when they stop being one. Every headcount number Finance reports, every dollar payroll moves, every access decision security enforces, resolves back to a record somebody in this function configured.

Same species as Finance’s ledger. Different subject.

And it’s the same sentence in every room. HR, IT, your CFO, your aunt at Thanksgiving. It needs no audience check, which is exactly why your router was never built to return it.

I read a manifesto last month from a services firm. Five points on what this function should become. Point one hands us the AI and agent strategy outright (not IT, not the vendor, us).

Read it twice and the interesting part isn’t the grant. It’s that they had to address it to “HRIS.” A system acronym. People arguing that we should own the most consequential decision in the building, and still no word for the people who’d own it.

They weren’t being careless. There isn’t one.

What I’d do with this Monday

Take the version of your answer you already use with people who understand the domain: not the family-friendly one, the real one. Write it down as a sentence with a verb and an owner: not “I do the Workday thing” but “I own [decision], I configure [system boundary], I’m accountable when [specific failure].”

That’s the sentence from the last issue’s exercise. Turns out most of you already have the raw material for it sitting in your own comment history. You just haven’t been asked to write the serious version down in one place.

Close

Ninety-one (and still counting) answers, and the most common one wasn’t a title. It was an if-statement.

The router is a good piece of engineering. Fast, accurate, and it’s kept this function employable through twenty years of nobody knowing where to file us.

It has one hard limitation. It only runs when a human asks.

Agents don’t ask. They read the record and act on it, at speed, without checking who’s in the room. The thing we never wrote down is about to be consumed by readers who can’t be charmed into a nod.

Stop naming yourself after the room. Start naming yourself after the record.

-Mike

Director HR Technology | Career Actuary

P.S. After I finished counting, I asked Justin what I do for work. No setup, no context. Just asked him at the table.

“You make sure people get paid right.”

One in ten of you gave that answer. He gave it in about two seconds, without checking who was in the room, because nobody’s taught him yet that there’s an org chart to hedge around.

Charlie Spitzer and a seventh grader, same category.

Fixed return value. No router required.

tldr;

The CEO of OpenAI says the singularity is here. I asked an AI, and it disagreed. So does my seventh grader.

Transformation never arrives as a rupture. It arrives as Tuesdays that accumulate (see: 33 years from Dr. Sbaitso to now, and the entire cloud era).

• Anyone selling you a "moment" is selling. The honest question isn't "when does it arrive," it's "what does this touch."

• Your governance backlog isn't competing with your AI strategy. It is your AI strategy.

• Homework for the quarter: run one AI feature in sandbox, break it on purpose, write down where it broke.

On Saturday, the CEO of OpenAI announced that humanity has entered the singularity.

He said this on a podcast, which feels right. The most consequential threshold in human history, the moment machines surpass their makers, revealed between ad reads.

So I did the obvious thing. I asked an AI whether it was true.

It said no.

Not a hedge. Not a "well, it depends on your definition." It told me (and I'm paraphrasing only slightly) that it can't improve itself, can't remember yesterday without a filing system bolted to its side, and that there's an entire building full of humans employed specifically to keep it pointed in the right direction. Its exact framing was "that's a powerful tool, not an event horizon."

The machine was more skeptical of the machine than the man selling the machine. Which is a bit like your realtor telling you the house is overpriced. You should probably listen, because that opinion is working against its own interests.

And yes, this was a real conversation. I did not invent a sassy robot for narrative convenience. (The robot came pre-sassed.)

The 33-year "moment"

My problem with singularity talk isn't that the technology is unimpressive. It's very impressive. My problem is the shape of the claim.

The singularity is a rupture story. A before and an after. One day the machines can't, the next day they can, and everything downstream is unrecognizable. It's a great structure for a movie trailer and a terrible description of how technology has ever actually arrived.

I know this because I've been watching this particular movie since the opening credits. In the early '90s I installed a Sound Blaster card, and bundled with it came a program called Dr. Sbaitso, a "psychologist" that talked in a flat robot voice and mostly repeated your own words back as questions. ("WHY DO YOU FEEL THAT WAY.") If you cursed at it enough, it threw a parity error and rebooted. I ran that experiment. Repeatedly. For science (Quick aside: Yes, I know ELIZA was first).

(Full disclosure: when I brought this up with the AI, I called it "Dr. Sebago." The AI corrected me, then observed that Dr. Sebago sounds like Dr. Sbaitso retired to Maine to sell boat shoes. So, let's review: The program from 1992 could not remember the sentence I had just typed. The program from 2026 fact-checked a memory I'd been mangling for three decades and roasted me about deck footwear in the same reply. That's the curve. I am living on it, and apparently so is my dignity.)

Dr. Sbaitso to the thing that argued with me this weekend is roughly 33 years. The jump is staggering. But notice what it wasn't: a moment. There was no rupture, no threshold, no day the sirens went off. There was a long, boring grind of incremental progress that only looks like magic when you compress three decades into a before-and-after slide.

There was no lunch table moment. There was just Tuesday. Over and over. For thirty-three years.

The kid who never said wow

Here's the part that actually convinced me, and it happened in my own kitchen.

I mentioned to Justin (12, seventh grade, currently running a 3D printing operation out of my basement with the margins of a small cartel) that the CEO of OpenAI says we've entered the singularity.

He looked up from slicing a print file and said, "What's a singularity?"

I explained. Machines smarter than people. The future becoming unpredictable. The thing every sci-fi movie warned us about.

He shrugged and said, "Isn't that just... what AI does?" And went back to his print file.

That shrug is the whole argument. Justin has used AI for homework help since he was ten. To him it's an appliance, somewhere between the microwave and the label maker in the household hierarchy of marvels. The rupture story requires a before, and Justin doesn't have one. You cannot sell "the moment everything changed" to someone for whom it was always Tuesday.

The singularity, it turns out, is a nostalgia product. It's only for sale to people old enough to remember cursing at Dr. Sbaitso.

You've already lived this movie

And if you work in the Workday ecosystem, you've lived it twice, because you remember when "the cloud" was going to be a moment.

Remember that? The revolution. The rupture…The day on-prem died… Bueller? Bueller?

What it actually was: a decade of migrations. Tenant by tenant, integration by integration, one white-knuckle cutover weekend at a time. The cloud revolution absolutely happened, and not one person who lived through it can point to the day. It arrived as a thousand Saturdays in a conference room with cold pizza, and by the time the analysts declared it complete, you'd been living in it for years.

AI in your tenant will arrive exactly the same way. No singularity release. No Monday morning where the platform has transcended. A feature here, an agent there, a capability quietly folded into a business process you already own. Tuesdays. Accumulating.

Which means the rupture framing isn't just wrong, it's operationally wrong. It trains you to wait for an announcement that will never come, instead of managing a curve you are already standing on.

What the curve means on Monday

Three things, practically.

• One: anyone selling you a moment is selling. When a roadmap slide promises transformation as an event (a version, a launch date, a keynote reveal with dramatic lighting), that's a marketing shape, not a technology shape. The honest question is never "when does it arrive." It's "what does this touch." Which data. Which security groups. Which business processes. Moments are for stages. Curves are for architects.

• Two: the boring stuff is the AI strategy. Every AI capability that lands in your tenant lands on something. Your data quality. Your security model. Your business process hygiene. Ship an AI feature into a tenant where the supervisory org is held together with duct tape and wishful thinking, and you will get duct-tape-and-wishful-thinking answers, delivered confidently and at scale. Your unglamorous governance backlog is not competing with your AI strategy. It is your AI strategy.

• Three: gradual doesn't mean ignorable. The trap on the other side. Because there's no rupture, it's tempting to conclude nothing is happening. But 33 years of Tuesdays turned Dr. Sbaitso into something I genuinely can't fully explain, and the people who dismissed the curve in 2023 are already explaining gaps in their resumes. Pick one concrete thing this quarter. Run an AI feature in sandbox. Break it on purpose. Write down where it broke. That's the whole assignment.

Where Altman has a point

Fairness requires this: he's not entirely wrong. He's wrong about the shape.

The compounding is real. And here's the uncomfortable part. Me, spending a Saturday casually debating philosophy with a machine and losing on points? That would have sounded exactly like the singularity to 1994 me, sitting in front of Dr. Sbaitso, waiting for the parity error.

The goalposts move because the middle of a curve never feels like the middle. Every generation of "wow" becomes table stakes in about eighteen months, and then we forget we ever said wow. Altman is looking at that curve and calling it a moment. I get the temptation. It's just not how anyone will actually experience it.

We'll experience it the way Justin already does. As an appliance. As a Tuesday.

The close

So no, the singularity has not arrived, and you don't have to take my word for it. Take the AI's. (Or my seventh grader's. They agree, which should terrify Sam Altman more than anything.)

But something is arriving, at the speed it always arrives: gradually, unevenly, and directly on top of whatever foundation you've built.

The singularity won't be announced on a podcast. It'll show up in your tenant on a Tuesday, disguised as a feature release. Be ready for that one.

— Mike

Director HR Technology | Still looking for the parity error

P.S. Yes, I went looking for a working copy of Dr. Sbaitso, because journalism. It's on the Internet Archive, running in a browser-based DOS emulator, which means a 1992 chatbot is currently enjoying better uptime than several enterprise systems I could name. I asked it about the singularity. It asked me why I feel that way. Thirty-three years on, still undefeated.

The Department of First Things First. For the people who do the work.

Fifty-five of you answered. Nobody agreed.

HRIS person. Workday admin. HR systems analyst. HR Tech. People Systems. Business systems. A few “depends who’s asking.” One of you said “professional translator,” which is the closest anybody got.

Twenty years in, and the most specific work in the building is done by people who can’t finish the sentence “I’m the—”

Here’s what I’ve been chewing on since. This isn’t a branding problem. It’s a scoping problem wearing a branding costume.

Every other function names itself by what it owns

Finance owns the ledger. Legal owns risk. Security owns the perimeter. Ask any of them what they do and you get a noun: a thing they’re accountable for, with a boundary around it.

Now listen to how we do it.

“Workday admin.” Named after a tool.

“HRIS analyst.” Named after a category of tool.

“I’m the one who fixes it when it breaks.” Named after a failure mode.

We name ourselves after the systems we operate or the disasters we absorb. Not one of those is a scope. They’re descriptions of proximity. I am near the thing. When the thing is bad, I am nearer.

The vendor-name trap

If your title is “Workday admin,” your professional identity is leased from someone else’s marketing department.

I’ve said before that vendor branding has a shelf life of about nine months, and I’ll keep saying it. Watch it happen in real time: Illuminate shipped with a naming architecture (Accelerate, Assist, Transform) that got quietly folded into Sana-based branding before most of us finished updating our slide decks. If you’d built your headline around any of those words, you’d be doing maintenance on your own identity right now.

The architecture is permanent. The branding is weather.

That logic runs all the way down to your title. Name yourself after a product and you inherit its release cycle. Name yourself after what you govern and you’re durable across a re-platform, an acquisition, and whatever the next three keynotes decide to call the same four ideas.

“I fix it when it breaks” is the most honest answer and the most expensive one

I love that answer. Every one of us has been that person at 11pm on a payroll weekend.

But it describes a service level, not a scope. And you cannot budget for a service level nobody ever wrote down.

That’s how a team of four ends up holding an entire ecosystem that Finance reads as one line item: because “the Workday thing” has never been described in terms of the decisions it controls. Only in terms of what it costs when it stops.

No name → no scope → no decision rights

This is the part that actually matters, and it’s why I’m not treating a poll about job titles as a bit.

A function with no stated scope doesn’t get decision rights. It gets consulted. There’s a difference, and you’ve felt it: you were in the meeting, you raised the concern, they thanked you, and the decision went the other way. You were the input. Somebody else was the owner.

Then it breaks in month fourteen and you get named after that, too.

Scope isn’t a title. Scope is a written answer to one question: which decisions cannot be made without me, and which ones are not mine to make.

Where the ambiguity is real

Fair concession, this job genuinely spans: Configuration, architecture, vendor management, change strategy, and occasionally light therapy. Any title narrow enough to be precise would be too narrow to be true.

And I know people who’ve worked the fuzziness on purpose. The undefined lane let them grow into finance systems, into analytics, into whole org design conversations that a clean title would have fenced off. That’s not nothing. Ambiguity has a growth setting.

But there’s a difference between choosing ambiguity and inheriting it. Most of us inherited it.

What I’d actually do Monday

Not fix your LinkedIn headline. That’s the cosmetic version.

Write the decision-rights list. One page, two columns. Column one: decisions that require my sign-off. Column two: decisions where I’m an input.

Then take it to your leader and get them to argue with it. Their edits are the whole point: Every line they move from column one to column two is a place you thought you had authority and didn’t. Better to find that out over coffee than in a postmortem.

Do that and the title question gets easy. You stop needing a name for what you touch, because you finally have one for what you hold.

Close

Fifty-five answers and no consensus isn’t a profession without a name.

It’s a profession that’s been describing where it stands instead of what it owns.

Stop naming yourself after the system. Start naming yourself after the decision.

tldr;

Nobody in this field chose it. We got matched to it, usually by a single keyword on a job posting.

Mine was SQL, in 2004. Within a couple of years I had upgraded an entire HRIS by myself, wired physical time clocks, and shipped plain-language HR analytics before the iPhone existed.

The sideways door isn’t random. It selects for exactly the people this work needs.

Also: the “revolutionary” conversational analytics your vendor demoed last quarter? I ran it in the mid-2000s. It was called EasyAsk. Hold that thought.

Reply with your origin story. I read every one.

In 2004 I was a temp.

SQL work for an e-discovery client. Paid the bills. Zero long-term plan, because I was in my twenties and the long-term plan was “keep having a job.”

Then I saw a posting. Somebody wanted strong SQL skills for something called an HRIS Specialist role.

I want to be honest about the decision process here. I did not feel a calling toward human resources technology. I did not have a vision of enterprise systems. I went to college for finance. I recognized one keyword. I applied because the ad said SQL and I spoke SQL. That’s the whole story. That’s everyone’s whole story.

I got the job (mainly because at the time, I was volunteering on a Linux project called MythTV and shipping code, and of course, SQL). And then the job got interesting.

Purple squirrel, first rodeo

If you’ve been in this field a while you know the term “purple squirrel”: the candidate who impossibly matches every requirement at once. Recruiters hunt them for years and never find one.

Here’s what nobody tells you: purple squirrels aren’t found. They’re manufactured, on the job, by small teams with big systems and no budget for specialists. You don’t match every requirement. You become every requirement, one emergency at a time.

At that first HRIS job I ended up doing everything, because there was nobody else to do it. I upgraded the HRIS itself (Personnel Data Systems) from an aging Windows 95 application to their web-based platform. Alone. Integrations included (lots of DTS and stored procedure work). I wired up Kronos time clocks, and I mean physically wired, as in “why is the HRIS guy holding a drill.”

And now the part that’s going to sound made up: I shipped conversational HR analytics before the iPhone existed.

I stood up an analytics package that let regular humans ask HR analytics questions in “plain language”. Type “what’s our headcount” in English, get an answer from the HR data. No report writer. No training class. Just ask.

The tool was called EasyAsk (holy cannoli, It's still around too). This was the mid-2000s. Flip phones. The natural-language query was “translated” (i.e. — mapped to SQL in the backend), ran against the HRIS and it worked, (mostly), which is roughly the same reliability rate as the version being demoed to your executives right now.

I need this audience to absorb that for a second.

“Ask your HR data questions in plain English” is the marquee demo of the 2026 AI era. Every vendor keynote. Every analyst deck. And a temp-turned-HRIS-guy was running it against a client-server HR system two decades ago, back when the revolutionary technology was a BlackBerry.

The technology underneath is different now, and genuinely better. But the lesson stands, and it’s the same lesson this newsletter beats weekly: the demo is never the hard part. The hard part in 2006 was the same as the hard part today. Whether the data underneath could survive being asked a question (and the mapping. Always with the mapping).

Everything old is new again. The people who’ve been here since the old version can tell you exactly where the new version is going to hurt.

The industry doesn’t let go

In 2008 the financial crisis (and a poorly timed back injury) did to me what it did to a lot of people. Laid off.

And here’s the detail I love in hindsight: two weeks after the layoff, I went to work for PDS. The company whose software I’d spent 4 years upgrading and wiring and interrogating from the customer side.

I didn’t escape the field during the worst job market of my lifetime. The field just moved me to the other side of the software. That’s the thing about HRIS: you don’t leave. The door only opens inward.

Twenty-plus years later I’ve been on the customer side, the vendor side, the implementation side, and whatever side “writes a newsletter about it” is. Every move traceable to one keyword on a job ad in 2004.

The sideways door is the point

There’s a podcast in our world, the HRIS Careers podcast, whose whole tagline is that no one gets into HRIS on purpose, but it sure is interesting to find out how they did. It’s the truest sentence anyone has said about this profession, and every origin story I’ve ever heard confirms it.

The person who “knew where the Access database lived.” The HR generalist who was “good with computers” and got handed the ATS. The payroll analyst who fixed one integration and was never allowed to do anything else again. The temp who knew SQL.

Nobody chose this. We got chosen, by proximity and competence, one weird problem at a time.

And here’s what I’ve come to believe about that: the sideways door isn’t a bug in how this profession staffs itself. It’s Ollivander’s shop (strap in Harry Potter fans).

The wand chooses the wizard. You walk in assuming you’re the customer, you wave a few sticks around, things explode, and then the right one finds you. And the old man’s explanation for why is the honest part: it’s not always clear. Not your transcript. Not your five-year plan. Something in the wiring.

That’s how this field recruits. Nobody picks HRIS off a shelf. The temp who knew SQL, the generalist who was good with computers, the payroll analyst who fixed one integration and was never allowed to do anything else again: the wand found every one of us, usually mid-explosion.

(Our ecosystem does contain one actual practicing wizard, the Workday Extend Archmage himself, Chris Humphrey, and I trust he’ll certify this framing as canon. The robes work because the magic is real.)

It only admits a specific kind of person. Someone curious enough to say yes to a job they can’t define. Someone unafraid of a system everyone else is afraid of. Someone who can stand between IT and HR and Finance and translate, all day, without belonging fully to any of them.

You can’t hire for that with a requisition. The wand has to choose you.

So no, we’re not failed IT people, and we’re not technical HR people, whatever that means. We’re the people who said yes to the weird job. Twenty years in, I’ve decided that’s not an accident to be embarrassed about. It’s the credential.

How’d you get in? Reply and tell me. I read every origin story, and I’ve never once heard a boring one.

— Mike

Director HR Tech | Purple Squirrel, First Rodeo

P.S. I told Justin I was writing about how nobody plans to end up in my field. He pointed out that he didn’t plan Justin’s 3D Lab either: “I just kept insisting you buy printers.” So there it is. Twenty years in HR Technology and the door still only opens sideways, even in my own basement.

The Department of First Things First. For the people who do the work.

No bullets today. Read the whole thing.

A few months back I published a breakdown of Workday's Flex Credits. The metered pricing. The grace period nobody reads correctly. The setup most teams still have not finished.

Practitioner stuff. The kind of thing I write so the people who build this for a living can walk into a budget meeting and not get caught flat.

It was neutral. Deliberately so. I do not sell Workday. I do not sell against Workday. I take the architecture apart and tell you what it does.

Then a reader forwarded me a sales email. From an account executive at a Workday competitor, going out to prospects.

And there, pasted into the pitch, was my work.

Not quoted. Lifted.

My headline, word for word. My subtitle, word for word. My summary. My grace period section, sentence for sentence. Dropped into sales material and aimed at buyers, to scare them off a platform I had analyzed without taking a side.

Neutral analysis went in one end. FUD came out the other.

And the targets were people on tight budgets, handed fear dressed up as my analysis and told to think twice. That is who it was built to spook.

Precision is the whole job

Whole passages, copied. Not a link with a pull quote. Identical down to the punctuation. I have the side by side.

They did not ask. They did not license it. They took a piece written to be vendor neutral and reframed it as a reason to fear one specific vendor.

Here is the part that got me. They did not even use my name. In the email I am "someone." I am "this person." They wanted the credibility of the analysis without the inconvenience of the byline.

They took the work and left the author in the parking lot.

I wrote about a pricing model. They turned it into a weapon. The fear was not in my work. They added it.

This was never really about the copyright

The copyright is real, and there are ways to handle that. I used them. Quietly. That part is done.

But the bigger point is bigger than one lifted article, and it belongs to all of you.

Independent practitioner writing only has value because it is independent. The entire reason you read a breakdown from someone like me instead of a vendor deck is that I am not running a play. The moment a vendor can strip-mine that work and bolt it onto a sales narrative, every independent voice in this space is worth a little less. Yours included.

Citation is a contract. You can quote me. You can argue with me. You can tell your customers I got it wrong.

What you cannot do is take my name off the analysis, take the neutrality out of it, and resell it as the case for your product.

That is not citation. That is laundering.

Watch your own bylines.

Your analysis is an asset. Treat it like one.

If a vendor ever quietly decides your independence is theirs to spend, notice it. Keep the evidence. Handle it. Quietly if you can, loudly if you have to.

I did not name the vendor here. Not because it is a secret, but because the lesson travels further than the grudge. Today it was my article. Tomorrow it is yours, and you find out from a forwarded email like I did.

I write about architecture over branding because branding has a shelf life and the architecture is permanent.

Turns out the same is true for a byline.

— Mike

P.S. I ran this past Justin, my in-house UX tester and resident seventh grader. Gave him the one-sentence version: someone copied my writing word for word, dropped it into their own thing, and never said it was mine.

He did not look up from his phone. "That's plagiarism. That's an automatic zero."

They teach this in elementary school. There is a rubric. There are consequences. My kid, who lost his volleyball shorts twice last month, knows you cite your sources.

One of us had a legal department. The other had a homework folder. Only one of them needed this explained.

tldr;

Two weeks ago the Bus Factor post said: you documented the what, and the why is still in your head. This is where the why goes instead.

Not a wiki. Not a documentation project. A ledger: four lines per decision, three minutes, written the day the decision happens.

The test for what earns an entry: if you’d have to explain it twice, it’s an entry.

The format: Date. Decision. Why. What we rejected. (Landmine, optional but recommended.)

The first copy of the why goes in the object’s own notes field (yes, the ones sitting blank in your tenant right now). The ledger is the commit log.

The audience is Month-14 you, audit-season you, your eventual successor, and the AI you will one day point at your tenant. In roughly that order.

I was dealing with an issue a few years ago that I want you to sit with: why does the offer approval route through a comp partner when the requisition is flagged remote? An HR Partner complained that it was a “useless routing step”.

Nobody knew.

So we did what every HRIS team does. Archaeology. Dug through old tickets. Searched Temas back three years. Found a thread that referenced a meeting that referenced a decision made by someone who left in 2019 (no notes, obviously, the meeting was the notes).

And then we did the other thing every HRIS team does. We kept the step. Because deleting a rule you don’t understand is scarier than living with one you can’t explain.

That’s how tenants calcify. Not through bad decisions. Through good decisions that outlived their reasons.

You cannot out-document the tenant

I know what the responsible voice in your head said after you ran the Bus Factor audit. “We need a documentation initiative.” Runbooks. A config workbook refresh. A wiki with a taxonomy and an owner and a quarterly review cadence.

Don’t.

I said it two weeks ago and I’ll say it harder now: a wiki nobody reads is just guilt with a URL. What-documentation is a photograph of a moving system. It’s stale the first time a release touches the config, which is to say, immediately. You will spend forty hours building it, feel wonderful for a month, and then it will depreciate in exactly the way the knowledge in your head does, except now there are two things rotting instead of one.

The problem was never the what. The tenant will tell anyone the what (that’s what a tenant is). The thing that dies when someone leaves is the why. And the why has a property that the what doesn’t:

The why never changes.

A decision happened once, on a specific day, for specific reasons, with specific alternatives on the table. That’s history. History doesn’t go stale. Which means you don’t document the system (unwinnable). You document the decisions (four lines, done forever).

The ledger

Here’s the entire practice. A running log, one place, append-only. Every entry:

DATE:      2026-03-11

DECISION: Comp eligibility rule excludes interns

WHY: 2019 acquisition brought 400 interns on a stipend structure that breaks the merit matrix

REJECTED:  Separate comp plan (Finance refused the headcount)

LANDMINE:  The intern flag feeds the benefits eligibility integration. Touch one, test the other.

That’s it. That entry takes three minutes to write and it just answered every question your successor, your auditor, and your own confused future self will ever ask about that rule. The Bus Factor post used this exact example as the knowledge trapped in your head. Now it’s trapped in a text file instead. Progress.

What earns an entry? Simple test: if you’d have to explain it twice, it’s an entry. The weird BP step. The security group shaped like a question mark. The integration you modified instead of replaced with an orchestration. The report you built because the delivered one lies. Anything where the config makes sense only if you were in the room.

Write it where the body is buried

Workday gave you a notes field on almost everything. Business process definitions have comments. Calc fields have descriptions. Condition rules, custom reports, integrations, security groups. Copious fields, shipped with the product, sitting blank in every tenant I have ever touched.

So the first copy of the why goes on the object itself. Paste the WHY line (dated) straight into the calc field description, the BP comment, the condition rule notes. Your successor doesn’t find it by knowing your ledger exists. They find it by opening the thing that confused them, which is the only place you can guarantee they’ll look.

Engineers figured this out fifty years ago: comments live in the code, history lives in the commit log. The note on the object is the comment. The ledger is the commit log. You want both, and here’s why the ledger doesn’t collapse into the notes fields:

Notes hold current state. People overwrite them. The reversals, the rejected alternatives, the story of how you got here: that’s the log’s job.

Cross-object decisions are homeless. The offer-routing decision touches a BP, an integration, and a security group. Which object gets the note? All three get a pointer. The decision itself lives in the ledger.

You can’t grep the tenant. Aggregating every comment field across every object type into one view is not a delivered experience. Your ledger is searchable today, from your phone, in one keystroke.

And this is where the robot argument gets teeth. When you eventually point an AI at your tenant, a dated why sitting in the object’s own description is context the model cannot miss. You’re not just keeping receipts anymore. You’re annotating the substrate.

The rules that keep it alive

Every documentation initiative you’ve ever watched die was killed by ceremony. So the ledger has exactly five rules, and four of them are about staying small.

Write it the day the decision happens. The why has a half-life of about a week. By Friday it’s fuzzy. By month three it’s vibes. By month fourteen it’s archaeology.

Two copies, both cheap. The why goes on the object, dated. The full entry goes in the ledger, which lives in one place (a doc, a channel, a text file; searchable beats structured). Thirty extra seconds. Worth it every time.

Three minutes, max. If an entry takes longer, you’re writing documentation again. Stop. Four lines. If it takes more than three minutes you will quit by August, and I’d rather have a scrappy ledger you keep than a beautiful one you abandoned.

No template committee. The moment this gets a SharePoint taxonomy, a review workflow, and a governance owner, it dies. It’s a ledger, not a program. Protect it from process the way you’d protect anything you actually need.

Append, don’t edit. When a decision gets reversed, that’s a new entry with its own why. Now you have the one thing no config workbook on earth contains: the story of how you got here.

Who you’re actually writing it for

Month-14 you. Seena Mojahedi’s framing from the last post: the SI leaves what-documentation because that’s what the SOW pays for, and around month fourteen something changes that nobody internal has the why for. The ledger is the thing the SOW never buys. Write it yourself, in real time, and Month 14 becomes a Tuesday.

Audit-season you. “Why is this configured this way” is the entire genre of auditor questions. Watching someone answer from a dated ledger entry instead of a shrug changes the temperature of the whole engagement.

Your successor. Someday there is one. The what they can read in the tenant. The ledger is the only artifact that makes them competent instead of merely credentialed.

And the robot. This is the part I need this audience to hear: sooner or later you are going to point an AI at your tenant. When that happens, the model will read your configuration fluently and explain it confidently, and without the why, some of what it confidently explains will be wrong (the config shows the intern exclusion; it does not show the acquisition, the stipend structure, or Finance saying no). Your ledger is the context layer. It’s the difference between an agent that knows why the rule exists and one that hallucinates a plausible reason and routes work accordingly. Clean your data before you buy the robot, yes. But also: keep your receipts, because the robot is going to ask.

Start with the audit

You don’t need to backfill a decade. You’ll never do it and the guilt will kill the habit.

Take the three or four spots the Bus Factor audit flagged (the ones where the honest answer to “who else knows this?” was nobody) and write those entries this week. From memory, dated today, flagged as reconstructed. Then keep the ledger going forward, three minutes at a time.

Your tenant is a decade of business decisions wearing a software costume.

Start keeping the receipts.

— Mike

Director HR Tech | Tenant Archaeologist

P.S. Justin keeps a log in Justin’s 3D Lab. Not of printer settings (the printer knows the settings). Of failures: what got messy, what he changed, why the fix worked. I asked him why he bothers. He said, “because I already printed the same mistake twice, and filament costs money.” A twelve-year-old independently derived the why ledger the second time a failure got expensive. Most of us are still waiting for the third or fourth.

The Department of First Things First. For the people who do the work.