Welcome to Field Notes -

Coming Soon

The Burning Building Nobody Sees: How the AI Revolution Gives Us One Last Chance to Build Work the Way It Should Have Been Built All Along

The foundational argument — drafting now, publishing 2026. Every week, one chapter’s core argument publishes as a Field Notes post on LinkedIn — written and audio. The site carries the full version. Field Notes is where the thinking develops in public.

FIELD NOTES: Issue no. 002

We Call Them Human Resources. We Treat Them Like Human Expenses.

The most dangerous assumption in business right now isn’t about AI. It’s about the people AI is being built on top of.

By David Hoke – May 2026

I got a call once from an associate on a borrowed cell phone.

They were sleeping in their car in the parking lot between shifts. They wanted to know the number for the EAP — the employee assistance program.

I gave them the number. And then I sat with the weight of what had just happened. Someone who worked for a company with a mission statement about people and values and family, calling from a borrowed phone, living out of their car, reaching for the one channel they’d been told existed.

That call never showed up in a single report.

The Gap Between What We Say and How We Build

We call them human resources. But if you watch how most organizations actually operate, we treat them like human expenses. Something to manage. Something to optimize. Something to reduce when the numbers get tight.

For thirty years, we’ve built entire systems around that assumption — cost controls, benefit programs, engagement surveys, wellness platforms — and asked them to carry a load they were never designed to carry. Meanwhile, the actual drivers of performance stayed untouched: schedules, workloads, policies, compensation design, operating pressure.

A four-thousand-dollar deductible on a fifteen-dollar-an-hour job. A shift that ends after the last bus leaves. A schedule that changes with forty-eight hours’ notice.

Those are not HR policies. Those are operating conditions. And operating conditions are the intervention.

The Numbers That Should Bother Every Leader

The data tells the story plainly. In 1970, roughly 38 workers died on the job every single day in America. By 2023, that number had dropped to 15 — with a workforce that had more than doubled in size. A 60% reduction in the death rate. That is what happens when you treat a problem as infrastructure: mandatory standards, shared accountability, data that everyone can see.

Now look at healthcare costs over the same period. In 1970, total US health spending was $74 billion. By 2024 it was $5.3 trillion. Not a gradual increase — a straight line up, decade after decade, through every wellness initiative, every People Experience rebranding, every Chief Wellbeing Officer appointment. The line never bent.

And engagement — the metric the wellbeing industry chose as its primary proof of value — just hit a ten-year low. After twenty-five years of measuring it, thirty-one percent of American workers describe themselves as engaged. The other sixty-nine percent? Somewhere between going through the motions and actively working against you.

Physical safety got infrastructure. Human performance got programs. One line went down. Two lines went up.

That is the burning building nobody sees.

The Outhouse Problem

I spent years buying outhouses. That’s not what I called it at the time. I called it building a wellbeing strategy. And I meant it — every vendor I brought in, every program I launched, every platform I integrated, I believed it would help.

Meditation apps. Step challenges. Digital therapeutics. EAP redesigns. Earned wage access. Biometric screenings. Financial wellness tools. I didn’t just approve these programs. I fought for them. I spent political capital to get them funded, spent weekends building the business cases.

I have held the Chief Wellbeing Officer title. I have sat in the chair. And what I learned is that the role, as it was designed across the industry, absorbed accountability without conferring authority. You owned the outcomes but not the inputs. You were responsible for employee health but couldn’t touch the organizational systems that determined it.

Here’s what I couldn’t see then: the building had no plumbing.

Every program I launched was sitting in the yard, disconnected from the structure itself. Inside the building, the operating conditions that actually determined whether people could perform and recover and sustain — scheduling practices, workload distribution, decision density, manager capability — none of those had pipes running to them. None of it was instrumented. None of it was engineered. We were installing increasingly sophisticated outhouses and congratulating ourselves on the upgrade while the building itself had no running water.

The Most Dangerous Assumption in Business Right Now

Here is where this becomes dangerous.

We are about to place the most powerful technology ever created — AI — into that exact system. A system that already treats people like a variable cost. A system that already optimizes for output over sustainability. And we are assuming the technology will carry the load.

It won’t.

Every era of economic transformation has had a moment like this one. The Industrial Revolution didn’t invent worker exploitation. It scaled it. The offshoring wave of the 1990s didn’t invent the employer-as-cost-center mindset. It accelerated it.

John Henry drove steel into rock faster than any man alive. They ran him against the steam drill — man versus machine. He won. Then he lay down and died. The steam drill kept going. It didn’t need to rest. It didn’t need water. It didn’t have a heart that could give out from the effort of being extraordinary under conditions designed to break it.

We are in a John Henry moment. AI is the steam drill. And the question isn’t whether your best people can keep up. Some of them will. The question is what happens to the operating conditions around them when the expectation becomes: keep up or get out.

AI will not fix broken operating conditions. It will scale them. Faster decisions. Tighter optimization. Less margin for human variability. If the system is wrong, it won’t drift. It will accelerate.

The Window

Here is the thing nobody is saying in the rooms where AI strategy gets made: the same technology that is accelerating the problem has also — for the first time in history — made the solution genuinely affordable and scalable.

The early workforce health interventions that actually worked — the ones proven in peer-reviewed trials at CalPERS and Bank of America — required extraordinary resources. Clinical stratification logic hand-coded in databases. Personalization engines that cost a fortune to maintain. The programs that bent the cost curve were expensive. Only the largest, best-funded employers could build them.

Today, a large language model handles that same logic in weeks. The personalization engine that once required a dedicated IT team now runs on infrastructure that costs pennies per interaction. We know what works. We proved it thirty years ago. We just couldn’t afford to do it at scale. Now we can.

But there’s a narrow window. The concrete is wet. Every AI workflow implementation being built right now is being poured on top of human infrastructure that was already under strain. The decisions being made in the next eighteen to twenty-four months will set for a decade.

Most leaders are walking up to this opening and looking through it at the technology. The ones who will build something worth building will turn around and look at the person standing behind them who has to operate it.

This isn’t a technology problem. It’s a design problem. We’ve spent decades trying to fix people instead of fixing the conditions they work inside. The window to get it right is open now. But windows close.

FIELD NOTES: Issue No. 001

The Architecture Problem: Why Employer Wellbeing Lost Its Way — and Why AI Forces a New Beginning

By David Hoke

Times they are a-changin’. Or maybe they aren’t.

Somewhere over the Pacific, flying home from Sydney, a thought surfaced that I couldn’t shake:

For all the progress we say we’ve made in workplace wellbeing, we’re still operating from the same architecture we built 30 years ago.

And that architecture was never designed for the world we live in now.

To understand why, we have to tell the truth about where this field came from — and what we quietly abandoned along the way.

In the beginning, wellness wasn’t about thriving. It was about costs.

When I first entered this field, the mandate was simple:

Help companies reduce healthcare spending.

That’s it.

Not engagement.

Not burnout.

Not human performance.

Not culture or belonging.

Just cost.

And for a brief moment, the strategy worked — not everywhere, and not for everyone, but in a very specific lane that we rarely talk about anymore.

The self-care era worked because it filled a gap the healthcare system doesn’t.

Before “wellness” became an industry, it was a set of targeted interventions created by researchers like Don Vickery, Jim Fries, and Kate Lorig.

Their insight was simple:

Most of health happens between doctor visits — and people need support there.

So they built programs that taught people with chronic conditions how to manage their lives, symptoms, medications, and decisions day to day.

The results were real, repeatable, and unusually strong:

15–17% fewer medical visits
ROI between 2:1 and 3.5:1
0.8 fewer hospital days per participant
Savings ~10x the cost of the program
Significant improvements in function, energy, and self-efficacy

These weren’t grand corporate initiatives.

They weren’t gamified or incentivized.

They didn’t rely on apps or wearables.

They were structured.

Human.

Educational.

And deeply aligned with the lived reality of chronic disease.

For a moment, we had an architecture that made sense:

support real people, with real needs, in real contexts — and outcomes improve.

Then the industry changed.

When wellness went mainstream, the evidence quietly disappeared.

As corporations adopted wellness, the focus shifted from targeted self-management to broad lifestyle risk reduction:

Weight loss.

Step goals.

Biometric screenings.

HRAs.

Challenges and incentives.

The promise remained the same:

reduce risk today → reduce costs tomorrow.

But the emerging evidence started to contradict that promise:

RAND: savings came almost entirely from disease management, not lifestyle programs.
University of Illinois RCT: no short-term impact on cost or health outcomes.
JAMA RCT at a national retailer: higher participation, no meaningful effect.

In hindsight, the problem seems obvious:

We attempted to engineer human behavior inside a one-year financial cycle.

We applied cost-containment tools to human-complexity problems.

The architecture cracked.

The uncomfortable question: What are we actually trying to accomplish?

In most companies today, wellbeing is expected to do everything at once:

reduce medical spend
improve engagement
address mental health
strengthen culture
reduce turnover
elevate human performance

But these outcomes do not come from the same system.

Cost containment relies on compliance and risk reduction.
Human performance relies on capacity, clarity, resilience, and belonging.

Trying to achieve both with the same set of tools is exactly why results rarely match intentions.

As an engineer once told me:

“Every system is perfectly designed to get the results it gets.”

And the system we built was designed for cost control — not human performance.

Which is why it still produces… cost control, at best.

And disengagement, at worst.

Then AI arrived — and quietly raised the stakes for everyone.

This is the pivot the industry has not fully absorbed.

AI will automate tasks.

AI will accelerate decision cycles.

AI will reshape what work feels like and demands from us.

But ironically, AI makes human wellbeing more important, not less.

Because the value humans now bring — the value AI cannot replicate — includes:

judgment
creativity
emotional intelligence
resilience
adaptability
collaboration
strategic imagination

Logging 10,000 steps does not build these capacities.

A screening does not build emotional intelligence.

A points program does not build resilience.

These grow when a person’s life quality improves:

Sleep
Mental clarity
Focus
Recovery
Stability
Purpose
Relationship quality
Psychological safety
Environment

In an AI-driven world, human performance IS the competitive advantage.

And human performance is built on life quality, not wellness programming.

This is the architectural shift we’ve been avoiding.

So where do we go from here?

We can’t fix this by adding more features or more point solutions.

We can’t fix it by tweaking incentives.

We can’t fix it with campaigns, portals, or quarterly challenges.

We fix it by rebuilding the architecture.

One that aligns with:

what humans actually need
what the science actually supports
what work actually demands
what AI actually changes

An architecture where human performance is the north star — and life quality is the operating system.

I’ll be writing more about this in the weeks ahead.

But for today, the point is simple:

Before we build the next system, we must tell the truth about the one we built.

It was designed for a different era — and that era is over.