When the workforce is half human and half AI, your 2018 board deck stops working. Workday is restructuring. Atlassian put it on the org chart. And the rest of us are about to find out who actually understands the flow of work and who was just researching it from a distance.

Another long one. Pour the coffee or spill the tea…..

Picture the room. CFO has the laptop open. Same KPI deck the company has used for fifteen years. Revenue. Gross margin. EBITDA. Headcount. CAC. LTV. Churn. Everyone is nodding because everyone has seen this slide a thousand times.

Then a board member with a tech background leans in and asks the question that breaks the room:

Crickets. Nobody has a clean answer. And that's the whole point of this newsletter.

The metric we have been tracking forever -productivity of a workforce - was built for a workforce made of people. Now the workforce is people and agents and the cost lines on the P&L are splitting in two and the HR tech platforms most companies depend on are being rebuilt mid-flight and the org chart itself is starting to move.

Three tectonic shifts at once. Most boards are reviewing dashboards designed for none of them. So let's get into it.

For seventy years, “labor” meant payroll. One column on the P&L, give or take a benefits accrual. Done.

In 2026, that column is fracturing. The Atlanta Fed's recent CFO survey (Atlanta Fed CFO Survey) found large US companies are now allocating about 55% of their AI spend to operations (subscriptions, services, training), 31% to internal development, and 13% to hardware. McKinsey's recent strategy work (McKinsey: The state of AI) makes the point sharper: AI's real impact will not show up as a productivity line item. It will reshape entire profit pools and force companies to rethink where competitive advantage lives.

Then there is the part nobody puts on a slide.

Microsoft, Amazon, Google, and Meta are spending a combined $725 billion on AI infrastructure in 2026, up roughly 77% year over year. Microsoft alone is at $190 billion. Meta is spending the equivalent of $370 million per day on data centers. Tech layoffs hit 81,747 in Q1 2026. Some trackers put the running total above 113,000 by early May, with one estimate at 864 people per day. (Capitaxer / Invezz analysis)

Alrighty, so this is FOUR companies with $725 billion in AI capex that is funded in part by laying off the humans the AI is supposed to be augmenting. The math is not mathing. It never was. And seriously – people cannot afford groceries, gas, electric bills right now guys – come on! Yeah – I am going there very soon. Stay tuned to my Letterman-esque top 10 FUGLY's coming soon.

Here is what every board needs to internalize. Your “cost of labor” line is becoming “cost of work,” and the cost of work is now three different cost curves stapled together:

•       Human labor - salaries, benefits, the people on your team. Scales with headcount.

•       AI labor - model API calls, agent runtime, fine-tuning costs, vendor licenses. Scales with usage. (And usage scales fast. Your invoice will surprise you and consumer electric bills…..)

•       Hybrid labor - humans supervising, correcting, and orchestrating AI output. Scales with complexity. Most underestimated category in the modern workplace and I will die on this hill.

Most companies still report this as one line. They aren't the same thing. They never will be again. This is a workforce turned upside down.

And here is the kicker nobody wants to say out loud: more AI usage almost always means higher costs in the short and medium term, not lower. The savings are theoretical and the invoices are real and please god do the math before you greenlight the next pilot.

Before we talk about how things change, let's name what's on the wall right now. Across NetSuite's 2026 financial KPI roundup (NetSuite KPI guide), ClearPoint's Balanced Scorecard work, and SEC disclosure guidance, US boards consistently track a small core set:

Financial: Revenue, revenue growth, gross margin, operating margin, EBITDA, net income, free cash flow, return on invested capital.

Customer: CAC, LTV, net revenue retention, churn rate, NPS, customer satisfaction.

Operational: Cost per unit, productivity per employee, on-time delivery, cycle time, utilization rate.

People: Headcount, voluntary turnover, employee engagement, time-to-hire, cost-per-hire.

Risk and governance: Cybersecurity posture (mean time to detect, mean time to respond), compliance readiness etc…

David Parmenter, who has spent his career on this question, argues boards should see only 9 to 12 measures, and most should be key result indicators (KRIs) rather than operational KPIs (Parmenter on KPIs). The principle is right. The list is also about to be rewritten by yours truly…. Brace yourselves.

(a.k.a. the dashboard everybody needs and nobody has built yet likely because the data sucks)

Every metric in section three carries a hidden assumption: the input being measured is human effort and when you pull that assumption out and watch the metrics fall over.

Productivity per employee becomes productivity per dollar of work. When an AI agent does in 4 minutes what a person did in 4 hours, productivity-per-employee goes parabolic and stops carrying meaning. The honest measure is output per dollar spent on work, where work includes humans, AI, and the orchestration cost between them.

Cost per unit becomes cost per outcome. Manufacturing has measured cost per unit since the dawn of the assembly line. In a knowledge-work economy where AI handles routine production, the unit gets fuzzy. What matters to the customer is whether they got the outcome they paid for, at what total cost, and how reliably. Boards will increasingly want cost per resolved customer issue, cost per qualified lead, cost per accurate forecast — not cost per hour worked unless there is a lot of overtime in there.

Headcount stops describing what it used to. A team of 8 people running 40 AI agents is not a team of 8. It is an operating capacity that needs a different name. Some boards have started asking for “effective headcount” or “augmented FTE” reporting. Expect that to become standard within two years. Revenue per employee will likely get replaced with revenue per AI or shall I say a blended model?

Margin expansion will need a footnote. Gross margins are going to look incredible at companies that automate aggressively. The McKinsey research is the warning label: in industries where competitors are doing the same thing, those productivity gains tend to get competed away into customer pricing. Boards that don't ask “is this margin durable?” will be surprised when it isn't. Don't be that board.

Churn becomes a relationship measure, more than a service measure. When AI handles the first 80% of customer interactions, churn isn't about whether the chatbot answered the question. It is about whether the relationship still feels human enough that customers stay loyal when something goes wrong. The 2024 Edelman Trust Barometer (Edelman Trust Barometer) found 61% of people worldwide believe business leaders are deliberately misleading them. AI deployed badly accelerates that perception. AI deployed well, with humans in the right places, can become a differentiator.

And then the new KPIs. The ones nobody has on the deck yet but everyone will, two years from now, while pretending they always did:

•       Human-AI ratio by function (customer-facing vs. internal work)

•       AI cost per workflow, with trend lines over time, because surprise! it goes up.

•       Quality drift - how often AI output requires human correction, and whether that rate is improving or worsening and what is the difference of the cost.

•       Blast radius - financial, regulatory, and reputational exposure if a given AI system fails publicly. Cybersecurity boards already think this way. Every board needs to. And insurance companies need to catch up real quick – can you cover all this and the many other lawsuits out there – there is a trend and lots of rightful suing going on in most cases – not all. Can you handle this incredible weight?

•       Time-to-trust for new AI deployments - how long it takes employees and customers to actually rely on a new system instead of double-checking it

•       Trust dividend - the discount your company earns on every future change when stakeholders believed your past AI deployments were thoughtful. We should really measure this since literally no trust exists these days in the companies and governments running/ruining the world. Just saying…..

•       People-skill capacity of remaining human roles - judgment, empathy, escalation, ethical decision-making. The stuff AI cannot do. Yes, it's measurable. We've been measuring it for 25 years and it is very important here!

That last one carries more weight than most boards realize. We will come back to it.

Now we get to the part that should be on every CEO and CHRO's radar this quarter, and probably this week. The HR tech sector is in the middle of the most consequential restructuring it has seen in a decade. Global HR software startups raised over $3 billion in the first half of 2026 (Ellty HR Tech VC list). Round sizes are bigger. Deal counts are smaller. April 2026 alone saw $73.4 million flow into seven HR tech companies (HrFlow.ai April fundraising recap). The capital is concentrating, and the thesis behind it has shifted from “more tools” to “which platforms understand the actual flow of work.”

The Workday story is the canary in the coal mine…maybe - lots of views on this one.

In February 2025, Workday cut 1,750 jobs (8.5% of the workforce) citing the need to prioritize “innovation investments like AI” (Fortune). Twelve months later, in February 2026, they cut another 400 — this round concentrated almost entirely in Global Customer Operations, with $135 million in restructuring charges (HR Executive). Two layoffs inside one calendar year. The second hit the team responsible for helping customers actually use the software. Yeah, read that twice – we need people helping here – no shit. AI is for the privileged – like 90% of the world still doesn't know – ok?

Then the CEO seat changed hands. Carl Eschenbach stepped down in late January 2026 after 13 months as sole CEO. Co-founder Aneel Bhusri returned with a public framing of “AI is a bigger transformation than SaaS” and a clear directive to pivot the company toward Illuminate, Workday's AI platform (KORE1 deep dive). Workday's stock has fallen roughly 40% over the past year. Investors are pricing in real concern that traditional enterprise HR software may not survive the agentic AI era in its current shape.

This is the canary. When the platform that manages your people starts cutting the people who help your team use the platform, you have to ask what is actually being optimized for. Margin? Sure. The customer? Doesn't look like it. The employee? Definitely not. And for the record, I have bought, endured and thrown out tons of HR tech platforms in my career. They all have 1 thing in common – no one likes them so start building with the customer and then maybe your margins won't suck.

The acquisition wave

Through the first eight weeks of 2026, the HR tech sector saw five notable acquisitions (UNLEASH):

•       Payoneer bought Boundless

•       Remote acquired Atlas

•       Phenom picked up Be Applied and Included AI (for the record, I like Phenom and they are also Philly – so even better!)

•       Docebo bought 365Talents

•       Perceptyx acquired Lyceum

Add Workday's pending acquisition of Paradox (which had raised $250M+ before the deal) and Workday's earlier $1.1B purchase of Sana, and the pattern becomes obvious: the legacy platforms are buying their way into AI-native capability because they cannot build it fast enough internally. They had a decade of head start. They blew it researching the future instead of building it.

Industry analyst Josh Bersin said it best — the bigger vendors are under pricing pressure from specialized AI-centric firms, which makes this a strong negotiation window for HR tech buyers (Bersin via UNLEASH). CHROs reading this: if your renewal is up in the next twelve months, this is your year to push. Hard. Don't sign the auto-renewal until you have priced two AI-native alternatives. In fact – tell all your vendors right now to take out the damn auto-renewal clause and start exploring. Own the tech and you own the decisions. NO MORE AUTO RENEWS PLEASE AND THANK YOU! P.S - make sure you are asking about pricing and Ai utilization rate too and if this is included in SaaS costs or extra…

What the investors are actually betting on

Read the recent VC commentary (Quartz: Investors pick their AI winners and losers for 2026) and one phrase keeps surfacing: “the gap between proof-of-concept and production is where money flows now.” In HR tech specifically, investors are betting on three things:

•       Agentic platforms that orchestrate end-to-end workflows rather than pipe data between dashboards

•       Frontline workforce optimization - platforms serving healthcare, retail, hospitality, and other sectors where shift-based work is the unit (Sona just raised on this thesis)

•       Skills, mobility, and hiring infrastructure that runs on real-time signal, not annual surveys

The losing thesis: generic AI products without a clear reason to exist. Look-a-like apps built on foundation models, chatbots without a use case, dashboards without orchestration, monolithic point solutions, and the god awful aged everything solutions – cough – workday – cough – cough. Crunchbase's 2026 forecast (Crunchbase) says it bluntly: it will be very difficult for a SaaS company without native AI/agentic capabilities to find VC dollars at any stage. Translation: if your HR tech vendor is still calling their AI “intelligent automation” and showing you the same product they showed you in 2022, they are dead and don't know it yet. A lot of them are going to die and probably should as we are all just AI wrapping some hot pink lipstick on an ugly pig.

The investment-execution gap (this should make every CHRO pause)

The Hackett Group's 2026 HR Key Issues Study (Hackett Group) found HR technology investment is up roughly 9% year over year. Staffing and operating budgets? Nearly flat. And 59% of HR organizations lack a clear implementation road map. The money is moving. The operating model is not.

The same study projected AI adoption in HR talent and performance review reaching 42% by close of 2026. Useful, but skewed toward transactions rather than the harder work of skills architecture and workforce planning. Translation: we are bolting AI onto whatever process happened to exist when AI showed up, instead of asking whether the process should exist at all.

“The organizations that will lead are those willing to reimagine how work gets done, not just adding new tools to legacy processes.” — Amanda Newfield, The Hackett Group

The winners in this market will not be the platforms with the most features. They will be the ones that understand the actual flow of work — what people need, when they need it, where the friction lives — instead of researching it from a corporate ivory tower and shipping a dashboard about it. That distinction is going to separate billion-dollar exits from cautionary slide decks at the next SaaStr conference.

Now to the announcement that, in my view, is the most important leadership move of 2026. Bigger than any product keynote. Bigger than any model launch. Bigger than the things that Elon keeps on shooting over my house.

Atlassian re-titled their CHRO. Avani Prabhakar is now Chief People AND AI Enablement Officer (Atlassian Leadership). Boom, about time – let's celebrate!

The People function and the AI function are now the same job. The CHRO owns the human side of the tech stack, end to end. Hallelujah and thank fucking god.

Why this matters: every major AI investment thesis right now bets on three things — workflow automation, agentic infrastructure, and human-AI collaboration. The CHRO sits at the center of it all. The companies that win will be the ones whose people leadership owns the technology stack their workforce uses, governs, and trusts. Atlassian just put it on the org chart. In plain language. Out loud. Who is next – LFG!

If you are a CEO or founder reading this, the question is no longer whether to elevate the CHRO into AI strategy. It is how fast you can do it without breaking the rest of your operating model.

If you are a CHRO reading this, the question is whether you are equipped for the seat that is about to open up at the table. And if you aren't, what's your plan to get there by Q3?

The KPIs that have to come with the new title

If your CHRO is owning the AI strategy, the board cannot keep reviewing the same People KPIs it reviewed in 2018. Here is the dashboard that needs to come with the new mandate:

•       Human-AI ratio by function - broken out customer-facing vs. internal, with a point of view on direction of travel

•       AI adoption depth, not breadth - not how many people logged in, but how many are using AI to do work that used to take a meeting

•       Quality drift and blast radius - the rate AI output needs human correction, and the exposure if a system fails publicly

•       Time-to-trust - how long until employees and customers actually rely on a new AI deployment and what this means with AI voice too. Yes, that is going to be a game changer or breaking.

•       Trust dividend - the discount you earn on every future change when stakeholders believed past AI deployments were thoughtful

•       Coaching and development spend - tracked the way R&D is tracked, because it determines whether the AI investment pays back

•       Trust behaviors — credibility, reliability, intimacy, and self-orientation. Four ingredients of trust from The Trusted Advisor (Maister, Green, Galford). Measurable for 25 years. They describe exactly what AI cannot do and yes I am associated with this company too – Relate – because it is cool and human and a good AI to make us be human again.

Most AI rollouts I see fail for the same reason most ERP rollouts failed in the 2000s. Leaders treat them as technology projects when they are behavior projects.

The math is simple. If you deploy an AI system that could deliver a 30% productivity gain, but only 40% of your team adopts it correctly, your realized gain is 12%. The other 18% is sitting in a slide deck somewhere with a really nice cover image and one messed up point in time change management plan and if you actually used it.

Adoption is a behavioral problem. So is trust. So is the willingness of a senior employee to admit they don't know how to use a new tool, or the willingness of a manager to coach a junior team member whose entry-level work is now done by an agent.

Change management for the AI period of work looks different than the change management of the past decade. It is less about communicating the why and more about:

•       Redefining what good looks like in every role, because the job itself is changing along with the tools

•       Coaching managers through the awkward middle, when AI does some of the work but humans still own the outcome

•       Building feedback loops fast enough to keep up, because AI capability is changing month to month, not year to year

•       Protecting psychological safety, so people will tell you when AI is producing bad output instead of rubber-stamping it

Here's the part of the conversation most strategy decks skip.

When AI handles routine cognitive work, the human work that remains is concentrated in the places AI is worst: judgment under uncertainty, ethical decision-making, persuasion, empathy, navigating organizational politics, building trust with people who matter, and knowing when not to use AI.

These have been called soft skills for so long that we underestimated them. The economic data is starting to flip the framing. McKinsey's recent work argues that competitive advantage will accrue to companies where “trust, judgment, and human accountability remain decisive.” The skills we have called soft are becoming the moat.

So for boards, the leadership pipeline metrics need a refresh. Years of experience is a worse predictor than ever. Demonstrated judgment in ambiguity is a better one — and harder to measure, which is why most companies don't even try. Performance reviews need to evaluate AI orchestration alongside individual output. Coaching and development spend should be tracked the way R&D is tracked. And trust behaviors need to be measured against a real framework, not vibes.

This section is for the people in between the executive team and the org chart. The people watching their PTO get cut. Watching their parental leave shrink. Watching their childcare benefit get “restructured.” Watching new hires get pushed into 996 schedules (9am to 9pm, six days a week, the imported-from-Silicon-Valley flex of working yourself into the ground for someone else's IPO). Watching their job description quietly absorb the work of two people who used to sit beside them.

Here is what nobody at the top is delivering clearly: some of your work will be automated. Whether you control that transition, or whether it happens to you, is the only real question. And the answer depends on what you do this year. Not next year. This one.

So here is what we do.

We become modern-day Luddites. Not the cartoon version. The real one.

The original Luddites were not anti-technology. They were skilled textile workers in 19th-century England who watched factory owners use new machines to slash wages, gut craft, and pocket the difference. They smashed the machines because the deal was rigged. They wanted technology that served the worker. They got fired, deported, and in a few cases hanged for the privilege. History remembered them as backwards. History was wrong. They were the canary.

A modern-day Luddite is not anti-AI. A modern-day Luddite is anti-rigged-deal. We use AI as a power tool. We learn it cold. We audit its outputs. We know when to put it down. And we refuse to accept that productivity gains belong only to the people who own the platform.

Upskill yourself. Now. Do not wait for your employer to do it for you.

Take the courses. Use the tools. Build something. Get fluent in the AI systems most likely to touch your function. Not as a passive user. As someone who can direct the tool, audit its outputs, and call it out when it is wrong. Use your brain, then automate your brain. Not the other way around.

Workers with AI skills are commanding wage premiums up to 56% above their peers (PwC Global AI Jobs Barometer). The premium is real and the window for capturing it is now.

Organize. Yes, I said it. Again.

If your employer is rolling out AI in ways that touch your hours, wages, surveillance, evaluation, or job security, you have power you may not be using. The Communications Workers of America negotiated the first AI-specific contract with Microsoft (Center for American Progress). The Teamsters won review rights over “meaningful technological change” in their UPS contract. The Las Vegas Culinary Workers Union won bargaining rights before AI deployment. The International Longshoremen's Association banned fully automated technology outright. Connecticut Senate Bill 435 would make AI use a mandatory bargaining subject for public-sector workers (Yankee Institute on CT SB 435).

Public approval of unions in the United States is at its highest level in half a century (AFL-CIO). The reason is not nostalgia. It is math. When the gains from technology get captured entirely by the people who already have capital, the people who don't have capital eventually find each other.

Not every workplace needs a union. Every workplace needs workers who understand they have the right to organize one. Same position labor took during industrialization. Same position during the rise of computing. Same position now. The pace is faster. The stakes are higher.

Protect your basic human rights. Loudly.

PTO is being shaved. Parental leave is being shortened. Childcare benefits are being “re-evaluated.” RTO mandates are being layered on top of 996 schedules. New mom and dad leaves are being quietly clawed back, with HR memos full of “market adjustments” and “realignment language.”

None of this is necessary. All of this is a choice. Every cut to a working parent's leave is a deliberate executive decision wrapped in a passive voice press release.

Document it. Compare notes with coworkers. Talk to your state representatives. Look up your state's pending labor legislation. Send the email. Sign the petition. Tell your CHRO directly when something they rolled out is hurting your team. Make the executives who decided your kid's care no longer fits in the budget have to put their name on the decision.

Build. Innovate. Create. Refuse to wait for permission.

Some of the most important new HR tech, mental health platforms, mutual aid networks, and cooperative business models of the next decade will get built by people who got laid off from a Workday or a Salesforce or a SkillCycle and decided not to wait for the next “professional CEO” to give them permission or doesn't even understand HR. The bottoms-up wave is already happening. Joining it is a choice.

•       Workday's Q4 fiscal-year earnings and whether Bhusri's Illuminate pivot is showing real customer pull or just press-release confidence

•       The next wave of HR tech M&A. Five acquisitions in eight weeks was the appetizer. Watch for two more big-vendor buys before Q3.

•       Atlassian's CHRO+AI title becoming a trend. If two more S&P 500 companies make the same org-chart move by year-end, it is no longer a one-off and the rest of the market will scramble to follow.

•       State-level AI bargaining laws. Connecticut SB 435 is the one to watch. Whichever framework passes first becomes the federal template.

•       The investment-execution gap closing or widening. If 59% of HR orgs still lack a clear AI road map by Q4, the layoffs in customer ops at platforms like Workday will start to look like a rounding error.

If you are a CEO or founder: read this newsletter to your leadership team this week. Then ask yourself — is your CHRO in the room when AI vendor decisions are being made, or do they get the calendar invite after procurement signs? If it's the second one, fix it before Friday please and quit being an asshole.

If you are a CHRO: stop letting Legal write your AI policy. Get in the room. Build the dashboard. Demand the seat Atlassian just put on the org chart. Your move is now. LFG!

If you are an employee: pick one AI skill, learn it cold, and tell your manager about it on Monday. Find out what your company's AI policy actually says. If you cannot find one, that is your answer. And if your benefits, hours, or leave just got cut in the name of “realignment,” send this newsletter to three coworkers. Then read up on your state's labor laws.

If this hit something for you: forward it to one person who needs to read it. Not five people. One person. The one whose face you saw when you read section nine.

If you got this from a friend: subscribe. The conversations we are having here are the ones nobody else is willing to have out loud.

See you in a week or so. The modern-day Luddite piece is queued up next — deeper dive on what we owe each other in this transition, and how the people who got stomped by the last wave of platform consolidation are quietly building the next one.

Best,

Kristy

SOURCES CITED IN THIS ISSUE

Cost line and AI capex: Atlanta Fed CFO Survey; McKinsey: The State of AI; Capitaxer / Invezz on Big Tech AI capex

Board KPIs: NetSuite Financial KPI Guide; David Parmenter on KPIs vs KRIs

Trust data: Edelman Trust Barometer 2024

Workday restructuring: Fortune (Feb 2025 layoffs); HR Executive (Feb 2026 layoffs); KORE1 deep dive on Workday cuts and CEO change