When leaders ask me where to start with AI transformation, they usually mean: which tools, which use cases, which teams first. Those are reasonable questions. But they're the second questions.

The first question — the one most organizations skip entirely — is this: what does your team's work actually look like at the task level?

Not the job description. Not the workflow diagram. The real daily work — what people are doing between meetings, the tasks they carry home, the things that keep getting bumped because something else always feels more urgent.

Starting there — before the tools, before the training, before the use case library — is the most direct path into transformation that actually changes how people work.

It's also the shift the research made. When economists first tried to estimate how AI would land on work, they scored whole occupations, and the picture came out either alarming or useless. The useful work only started when they broke jobs into their individual tasks.

In the study that set the standard, researchers found that about 80 percent of US workers have at least 10 percent of their tasks exposed to large language models — meaning the task could be done significantly faster at the same quality — and roughly one in five workers could see that for at least half of what they do.

What struck me most was the next number. On the raw model alone, the researchers estimated that around 15 percent of all work tasks could be sped up that way.

But once you account for the software and tools built on top of the model, that share jumps to somewhere between 47 and 56 percent — close to half of all tasks.

It's worth noticing when this was measured. The study came out in early 2023, near the very beginning of this technology's arc, and the models have advanced enormously since. If these estimates are off today, they are almost certainly low.

And none of those numbers mean anything until you do the thing the researchers did: break the work into tasks and look at it one task at a time.

I Have Done This Before

The first time I did this work, it wasn't called AI transformation. It was called digital transformation, and the tools were nothing like what we have now.

Years ago at VMware, I built a five-year digital transformation roadmap. Before I could build any of it, I did something that felt almost too simple to count as strategy: I sat with people while they worked, and I wrote down what they were doing.

Not what their job description said, not what the process map claimed — what they actually did, task by task, hour by hour. That was how we found the openings, the places where something repetitive and rules-based could become digital and give people their time back for the work that needed a human.

People were nervous then too. They wondered whether the technology would make their skills matter less. It's the same question I hear now, just wearing different clothes.

And the answer turned out to be the opposite of what they feared: the people who got time back got to do more of the work only they could do.

What's changed is what we're looking for. Back then, we were looking for "if this, then that" — the deterministic, repeatable steps a system could follow without judgment. That was what could be automated, and not much else.

AI moved that line. Now we're looking for repeatable judgment — the places where someone makes the same kind of call over and over, where their expertise has a pattern even if no one has ever written it down.

That's a far larger territory, and a far less visible one. Which is exactly why you still have to look.

Use Cases Are a Starting Point, Not a Strategy

Most AI adoption in organizations starts with examples — and that's not a bad thing. Someone demos AI answering emails. Someone shares a prompt that generates a project summary in thirty seconds.

A use case circulates in a team meeting, people try it, and some of them discover that AI is less intimidating than they thought. That's real value. Use cases lower the barrier and give people a first foothold.

The limitation shows up in the next step. When someone tries to apply another person's use case to their own work and it doesn't produce the same results, doubt sets in — and doubt is expensive.

People who came in hoping to save time, think at a higher level, or make a real dent in their workload walk away wondering if AI is actually worth the effort.

A field experiment with several hundred consultants at Harvard and BCG showed exactly why this happens. On the tasks that sat inside what AI does well, the consultants got dramatically better. But on tasks that fell just outside its range — and the boundary is jagged, not a clean line — the ones using AI did meaningfully worse than the ones working without it, by around 19 percentage points.

The tool didn't change. Whether it helped depended entirely on whether the task happened to fall inside its frontier, and that frontier sits in a different place for every kind of work.

That experiment ran in 2023 and was revised in 2026, and the frontier it mapped has not stood still since — it moves every time the models improve. Which is the deeper reason a borrowed use case ages so quickly: the boundary it was built around has already shifted.

The reason underneath this is that knowledge work is genuinely variable, person to person, task to task. The way one person preps for a board presentation, manages a stakeholder relationship, or synthesizes a set of research findings is different from anyone else doing the same job title at the same company.

One of the largest studies of generative AI at work makes the point in numbers: customer support agents using an AI assistant resolved about 14 percent more issues per hour on average, but the newer workers improved by 34 percent while the most experienced barely changed at all.

Same tool, same role, completely different results depending on who was holding it and their existing knowledge base.

Someone else's use case can show you what's possible — but it can't show you where AI fits in your work. That requires actually looking at your work.

That skill has to be developed — and the journal is how you develop it.

Two Levels of the Same Work

There are two places this looking has to happen, and they're easy to confuse.

The first is the organizational level — the interconnected, documented processes that run across a team or a department. Mapping those well, going deep on one real workflow rather than skimming all of them, is its own discipline.

It's the work I wrote about in What Discovery Actually Requires: the stated process and the lived process are almost never the same thing, and you find the difference by going deep on one workflow, not by collecting a library of use cases.

This piece is about the other level — the individual one. Organizational workflows have a defined shape that can be mapped and redesigned. Personal workflows are different.

They're how each knowledge worker actually gets their work done: the patterns they've developed, the judgment calls they make without naming them, the way they move through a week. These are variable, non-routine, and almost never written down — and they're where a huge share of the individual AI opportunity lives.

Paul Daugherty and James Wilson call the space where humans and machines produce the biggest gains the missing middle, and it doesn't live in a process diagram. It lives in the daily work of individual people.

They're not written down for a reason older than AI. The philosopher Michael Polanyi put it as plainly as it can be put: we know more than we can tell.

The most valuable things a skilled person does are often the ones they can't fully explain — the experienced nurse who senses something is wrong before the chart shows it, the manager who knows which conversation to have first.

The documented process and the real one differ not because anyone is hiding anything, but because the real one lives in the doing, not the describing.

Here's the part I want to be careful about: these two levels are not in competition, and neither one replaces the other. A lot of the value organizations are getting from AI right now came from the bottom — from individuals.

In Microsoft's global survey, 75 percent of knowledge workers were already using AI at work, and 78 percent were bringing their own tools to do it, often with no training or strategy around them. That's real momentum.

But on its own it produces hundreds of people solving the same problems in isolation, with no shared learning and no way to protect what they're building.

And organizational strategy built without ever looking at how individuals actually work produces tools that are technically right and contextually wrong. You need both.

The Assignment

Before any AI transformation work begins with a team, I ask everyone to keep a detailed work journal for two weeks. It tracks two things.

The first is tasks — and the energy behind each one.

Not a rating system, but a real account. What did you work on? Did it give you energy or take it — was it the kind of work that puts you in a state of focus, or the kind that depletes you as you go?

This isn't about what's important or high-status. It's about what's actually happening inside the work.

Here's an example from my own workflow. I used to read AI articles, record voice memos with my reactions, paste both into a Google Doc, download it to my desktop, and repeat that cycle enough times to eventually write a Substack article.

I liked reading the articles — the visual experience of seeing a headline, deciding if a piece mattered. That part I genuinely enjoyed.

Everything after it — the copy-paste, the file management, the drafts stacking up with no clear way to rank them — drained me every time.

I thought the whole routine was "reading articles." Once I mapped it at the task level, I could see exactly where the friction was — and that became the blueprint for what I built with AI.

The second is the work that's not getting done.

This column is the one people underestimate. Every knowledge worker has a list of things that matter — the strategic thinking, the relationship building, the creative work, the long-horizon planning — that keeps getting bumped by whatever is urgent.

This isn't about what anyone has fallen behind on. It's about what they're trying to protect space for. That pile is data — it's telling you what's being sacrificed so the friction can continue.

Two weeks is not arbitrary. Knowledge work varies enough from week to week that a shorter window gives you a snapshot rather than a pattern. You need to see the full range before the real picture emerges.

What the Data Is Actually Telling You

The energy signal is not a rule. Not everything that energizes someone should stay human, and not everything draining should be handed to AI. The journal is not a sorting mechanism. It is a way of making the invisible visible.

The real-world data backs that up: when Anthropic studied millions of actual conversations with its AI, mapped against a database of around 20,000 specific work tasks, the most common pattern wasn't automation — it was augmentation, people working alongside the tool rather than handing the task over entirely.

And it has to be a journal, not a memory, because we are surprisingly bad at seeing our own work clearly. Ask someone what they did last week and you'll get the three or four things that stood out.

Have them write it down as it happens, and a different picture emerges: how many genuinely different tasks they moved through, how little of the week actually went to the work they'd name as their real job.

I wouldn't have known what drained me in my own routine without having looked. That is the whole reason the journal works: it shows you what reflection alone will not.

When someone has documented two weeks of actual work — the tasks, the texture, the things they never got to — they can walk into the discovery conversation and say something specific. Not "I feel overwhelmed" or "I think AI could help with email."

They can say: here is the shape of my week. Here is where the friction is. Here is what I am trying to reach that I have not been able to.

That is a different starting place entirely.

What Leaders Do With This

Your job is not to analyze the journal data for your team. It is to create the conditions where gathering it is worth their time.

That means being direct about what this is for. The journal is not a performance assessment. It is not a way of identifying who is or isn't productive.

It is the pre-work that makes transformation actually fit the people doing the work — rather than asking people to fit a transformation designed without them.

Tell your team what they discover will shape what you build — not the other way around.

Then ask three questions:

What about this process could be better for everyone involved?

What task in the last two weeks do you wish you hadn't had to do yourself?

What's the work you keep meaning to get to that keeps not happening?

Those questions, before anyone opens a tool or attends a training, are where the real transformation starts.

The fear your team is carrying about AI — whether their judgment will still matter, whether they'll be able to keep up, whether they're already the person who got left behind — doesn't go away when you hand them a capability.

It starts to shift when they experience that what they know about their own work is being taken seriously as the starting point. The journal is one way to demonstrate that.

Someone Still Has to Do This Work

Most jobs, as the research keeps showing, are full of AI-touchable tasks. But knowing which ones, and whether AI actually helps with them, and how to fit it to one person's real work — that isn't something a tool does for you. It takes expertise.

There's an expensive way to get that expertise, and it's the one most people reach for first: bring in consultants to study the work and re-engineer it.

I can't tell you how many companies are actually doing that — but I can tell you what it would cost, because the work is variable enough that someone has to look at it one role at a time, and that runs into the millions per department fast.

And here's what rarely makes the pitch: when all of it is done, you still need people. People to run it, to judge it, to catch where it's wrong, to keep adapting it as the work keeps changing.

There's a better version, and it's the whole reason I have people keep the journal themselves. The ability to look at your own work and find where AI fits isn't a one-time deliverable someone hands you.

It's a capability your people can build and keep — one that pays off again next quarter and next year, as the tools change again.

A company that teaches its staff to do this themselves isn't just saving the fee. It's building a workforce that can keep doing it. The journal is the first rep.

The Leaders Who Get This Right

This is not primarily a data exercise. It is a decision about where transformation starts.

Most organizations start with tools and work backward, trying to fit people to the capability. The pre-work journal inverts that.

It starts with the people — with what their work actually looks like, with what they're losing in the current design. The tools come later, shaped by what you found.

That requires a leader who is willing to know what's really going on. Not the polished version, not the capability deck version, but the real texture of how your team works and what it costs them.

The leaders who get AI transformation right won't be the ones who moved fastest. They'll be the ones who stopped long enough to actually look.

Sources

Tyna Eloundou, Sam Manning, Pamela Mishkin, Daniel Rock, "GPTs are GPTs: An Early Look at the Labor Market Impact Potential of Large Language Models" (2023)

Fabrizio Dell'Acqua et al., "Navigating the Jagged Technological Frontier" (Harvard Business School / BCG; Organization Science, 2026)

Erik Brynjolfsson, Danielle Li, Lindsey Raymond, "Generative AI at Work" (Quarterly Journal of Economics, 2025; NBER Working Paper 31161)

Microsoft and LinkedIn, 2024 Work Trend Index

Anthropic Economic Index

Michael Polanyi, The Tacit Dimension (1966)

Human + Machine (Updated and Expanded) by Paul R. Daugherty and H. James Wilson

Intentional Design: "What Discovery Actually Requires"

ManpowerGroup measured AI adoption across 19 countries last year. Usage went up 13%. Confidence went down 18%. At the same time, in the same organizations.

More people using AI. Fewer people feeling capable with it.

That's not a rollout problem. That's a measurement problem. Nobody was watching the right things.

Michael J. Jabbour, AI Innovation Officer at Microsoft's Office of the CTO, frames it plainly: data is what makes you outcome-oriented. It tells you whether your activities, your initiatives, your programs are actually producing the results you want — or just producing activity. In AI transformation, the results most organizations track can look exactly right while the human layer underneath moves in the wrong direction.

HBR found productivity and psychological engagement moving in opposite directions under AI adoption — output up, motivation down. Upwork found the most productive AI users are nearly twice as likely to report burnout and twice as likely to consider quitting. The people driving your transformation numbers may be your highest flight risk.

You cannot protect what you cannot see.

Here is what measuring the right things actually looks like.

The Human Layer Assessment is twenty questions and a 1–5 scale. Take it before transformation begins. Take it again in a month — or after 90 days of active AI use if you're mid-program. The gap between the two scores is your actual data — which cluster moved, which didn't, where to focus.

Three Clusters

The questions fall into three clusters.

Conditions — what needs to exist before transformation can take hold: buy-in, inner orientation, learning culture, relational quality of the team.

Outcomes — what transformation is doing to your people: whether capability is building, whether work still feels meaningful, whether the pace is sustainable.

Reach — whether what your people are learning is actually spreading to the team, through shared practices, human connection, and collective standards for what good work looks like.

One screening question first, on the same 1–5 scale:

How actively are you currently using AI tools in your work? 1 = Not at all yet → 5 = Central to how I work every day.

This isn't scored. It segments your data. A team of early-stage users tells you something different from a team of daily users. You need to see that distinction in the results.

Score each cluster separately. The pattern across all three tells you more than any single question.

Rate each statement 1–5: 1 = Strongly Disagree → 5 = Strongly Agree

How to Run It

Send it to your team and run it anonymously. When people don't feel safe being honest, they answer in a way that sounds right — not in a way that's true. Anonymity is what makes the data usable.

This runs best when HR or a People leader owns administration and a senior leader owns acting on the results. Without someone accountable for both, scores sit in a deck.

If you want a directional read before committing to all twenty questions, start with Q1, Q8, and Q13. Those three — taken together — tell you whether the conditions, the culture, and the capability are present. If all three score low, you know where to begin. If one scores low, you know which cluster to prioritize.

Include a role or function field — not a name. That one extra question lets you see whether different parts of the organization are having different experiences. Without it, you only get one number for everyone — and a team that's struggling won't show up.

Conditions

The Conditions cluster measures the foundation underneath everything else: whether people understand why AI is happening, whether they feel ownership over it, whether the environment supports learning, and whether the team trusts each other. When no one explains the real reason, people fill in the blank themselves — and right now, the default answer is replacement.

Buy-in and Agency

Q1. I understand why my organization is adopting AI — not just that it is.

HBR found employees perceive AI as a direct threat to their competence, autonomy, and sense of belonging — before a single tool is introduced. BCG found leaders believed 76% of their employees were enthusiastic about AI adoption. Actual enthusiasm was 31%.

That 45-point gap lives here. This question measures whether the foundation for genuine buy-in exists — not whether the announcement happened, but whether it landed.

Q2. I feel in control of how AI gets used in my work.

Agency is what separates people who build with AI from people who comply with it. As AI systems encode how you think and work, who controls that is genuinely unresolved. Microsoft's Work Trend Index found only 13% of workers feel rewarded for reinventing how they work with AI. Low scores here mean AI is being done to people. High scores mean they feel ownership over how it shows up in their work.

Inner Orientation

Q3. I believe my ability to work with AI will improve the more I practice it.

Growth mindset is the strongest predictor of persistence through difficulty. When tools change and learning gets hard, this is what keeps people moving forward. HBR found only 5% of employees qualify as sophisticated AI users. The belief that capability grows with practice is what gets people there.

Q4. Learning to use AI feels like something I'm choosing, not just something I have to do.

Intrinsic motivation determines depth of adoption. Someone who feels ownership over their AI learning explores, experiments, and finds real applications. Someone going through the motions uses AI the way it came out of the box. BCG found that with strong leadership support, employee enthusiasm rises from 15% to 55%. This measures the starting orientation that support either unlocks or can't overcome.

Q5. I can stay focused and calm when AI creates confusion or uncertainty in my work.

AI creates real uncertainty — outputs surprise, tools change, workflows break. HBR found cognitive overload from AI oversight measurably degrades performance. Whether someone can stay regulated in that moment — rather than spiraling or reverting — is what determines whether learning compounds. This measures the emotional foundation underneath everything else.

Learning Culture

Q6. My manager shares what they're learning about AI, including when things don't go as expected.

This is modeling — the visible, imperfect practice of learning in front of the team. It is not the same as development, which Q10 measures.

Microsoft's Work Trend Index found that when managers openly model their own AI learning — including failures — employee trust increases 30 points. When managers perform expertise instead, they signal that not-knowing is dangerous. That signal shuts down experimentation across the whole team before it starts.

Q7. I know where to turn when I get stuck with AI.

People who hit a wall with no path forward revert. Gallup found less than one-third of employees in AI-implementing organizations say their manager actively supports their AI use. This doesn't have to be the manager — it can be a champion, a peer, a resource. It measures whether the infrastructure for getting unstuck exists at all.

Relational Culture

Q8. I feel safe trying things with AI here, even when I'm not sure they'll work.

Psychological safety is the condition that makes everything else possible. Without it, people perform adoption rather than build it. MIT Technology Review found 83% of executives say psychological safety measurably improves AI initiative success. This measures whether that safety is felt at the individual level — not assumed from above.

Q9. My team trusts each other's judgment, even as AI changes how we work.

Deloitte found high-trust teams use AI at 83% versus 63% for lower-trust teams — a 20-point adoption gap driven entirely by the relational environment, not skill or access. Trust also keeps teams from fracturing as adoption speeds differ across the group.

Q10. My manager actively supports my development through this AI transition.

Development is what managers do for you — actively investing in your growth, treating this as a human development opportunity, not a technology deployment. That is different from modeling, which is what they do visibly in front of you. Both matter. They are not the same thing.

Microsoft found employees report significantly higher AI readiness when managers actively develop them — not just encourage them to use tools. Low scores here tell you the program is being treated as a rollout. That distinction changes the outcome.

Those are the conditions. What comes next is what transformation is doing to your people.

The Layer Nobody Else Is Measuring

The Outcomes cluster measures what transformation is actually doing to your people — whether capability is building, whether work still feels meaningful, whether the pace is sustainable.

Capability and Judgment

Q11. I know how to direct AI to get useful results — not just generic ones.

Before confidence, before calibration — can someone get AI to produce something specific and useful for their actual work? Low scores mean adoption is wide but shallow. People are using AI the way it came out of the box, getting generic outputs, and moving on.

Q12. I feel confident using AI to get results that actually help me do my job.

ManpowerGroup found usage up 13%, confidence down 18% — simultaneously. More use is not more capability. This is the leading indicator of whether adoption deepens over time or plateaus.

Q13. I trust my judgment to recognize when AI output is wrong or needs correction.

The human backstop. HBR found AI output passed downstream without human review affects 41% of knowledge workers and costs roughly two hours of rework per instance. Whether people trust their own judgment to catch AI errors is what separates practitioners from users. Watch this delta closely post-program.

Q14. I can tell when it makes sense to use AI versus when I should handle something myself.

Calibration is the core judgment skill of AI-era work. Over-delegating produces cognitive atrophy. Over-involving yourself when AI could handle it causes burnout. HBR found that because AI now handles the repetitive tasks that once built judgment, people miss the formative practice. This measures whether that calibration exists.

Role Identity

Q15. My work still feels meaningful, even as AI handles more tasks.

HBR found productivity and psychological engagement moving in opposite directions — output goes up, motivation goes down. Meaning sustains engagement beyond the initial excitement. A drop here from pre to post signals transformation producing output without building the human outcomes that make it last.

Q16. AI is expanding what I can contribute, not replacing the judgment my role requires.

There are two ways AI changes what people contribute. HBR found that when people use AI as a reasoning partner — thinking about their thinking, not just producing faster — it expands what they can contribute. When they don't, output goes up but the distinctiveness of their thinking declines. This question measures which direction AI is moving people.

Wellbeing

Q17. I manage my energy well when doing cognitively demanding work.

Upwork found the most productive AI users nearly twice as likely to report burnout and twice as likely to consider quitting. HBR found AI introduced continual attention-switching and a growing number of open tasks — less time, not more. This catches the sustainability signal before it shows up in attrition.

Individual Gains Stay Individual

The Reach cluster measures whether what your people are learning is actually spreading to the team — through shared practices, human connection, and collective standards for what good work looks like.

Q18. My team shares what's working — including what didn't — not just polished final results.

Microsoft's research found AI doing to collaboration what remote work did — removing the functional dependencies that once required people to reach out to each other. Individual AI gains are real. Without a sharing culture, they stay individual. This measures whether organizational learning is actually happening.

Q19. I seek out colleagues' perspectives even when I could get an answer another way.

HBR found 74% of knowledge workers now use AI for at least one form of support they used to get from colleagues. Ju and Aral at MIT found human-AI teams communicate significantly less social and emotional content over time — they produce more but build less together. This measures whether human connection is still being chosen. That choice is what keeps collaboration capital from depleting.

Q20. My team has clear standards for what good work looks like.

HBR found that without shared quality standards, AI output erodes trust between team members — people can't distinguish someone's real thinking from something passed through a tool unchecked. Microsoft found teams that discuss AI quality standards are significantly more likely to be in the high-performing group. Standards are what make collaboration sustainable as AI changes what the work looks like.

What the Scores Tell You

Average the responses within each cluster. Your result is still on a 1–5 scale. Score each cluster separately. These are signals, not verdicts. The point of this assessment isn't to grade your organization — it's to show you where to focus.

Low Conditions — focus here first, before investing more in tools or training. Everyone feels the urgency to move on AI. This score tells you that the foundation — the why, the safety to experiment, the trust in the team — needs attention before anything else will land the way you want it to.

Low Outcomes — transformation is producing activity without building the human results that make it sustainable. People may be using AI more and growing less. That's a work design problem, not a training problem.

Low Reach — individual gains are staying individual. What one person figured out isn't moving to the team. Champions, shared standards, and structured knowledge sharing are the levers.

Low across all three — start with Conditions. You cannot build Outcomes or Reach without the foundation.

High across all three post-program — your people came through more capable, more engaged, and more connected than when they started. That's the outcome worth designing for.

The specific work practices that move each of these scores — what leaders, managers, and teams can do differently — are covered in the next article in this series.

The Data Already Exists — Inside Your Team

You can run this today. Before any formal program. Before you've spent anything on transformation. With your existing team, on your existing tools.

Take it now. Take it again in a month — or after 90 days of active AI use if you're mid-program.

The organizations that will get AI transformation right aren't the ones with the biggest budgets. They're the ones who knew where their people started and where to pivot.

The Assessment

Screening question (unscored, 1–5): How actively are you currently using AI tools in your work? 1 = Not at all yet → 5 = Central to how I work every day

Rate each statement 1–5: 1 = Strongly Disagree → 5 = Strongly Agree

Conditions

Q1. I understand why my organization is adopting AI — not just that it is.

Q2. I feel in control of how AI gets used in my work.

Q3. I believe my ability to work with AI will improve the more I practice it.

Q4. Learning to use AI feels like something I'm choosing, not just something I have to do.

Q5. I can stay focused and calm when AI creates confusion or uncertainty in my work.

Q6. My manager shares what they're learning about AI, including when things don't go as expected.

Q7. I know where to turn when I get stuck with AI.

Q8. I feel safe trying things with AI here, even when I'm not sure they'll work.

Q9. My team trusts each other's judgment, even as AI changes how we work.

Q10. My manager actively supports my development through this AI transition.

Outcomes

Q11. I know how to direct AI to get useful results — not just generic ones.

Q12. I feel confident using AI to get results that actually help me do my job.

Q13. I trust my judgment to recognize when AI output is wrong or needs correction.

Q14. I can tell when it makes sense to use AI versus when I should handle something myself.

Q15. My work still feels meaningful, even as AI handles more tasks.

Q16. AI is expanding what I can contribute, not replacing the judgment my role requires.

Q17. I manage my energy well when doing cognitively demanding work.

Reach

Q18. My team shares what's working — including what didn't — not just polished final results.

Q19. I seek out colleagues' perspectives even when I could get an answer another way.

Q20. My team has clear standards for what good work looks like.

Sources

ManpowerGroup Global Talent Barometer 2026

HBR "Research: Gen AI Makes People More Productive — and Less Motivated"

HBR "Why Gen AI Feels So Threatening to Workers"

HBR "What the Best AI Users Do Differently"

HBR "When Using AI Leads to Brain Fry"

HBR "How Do Workers Develop Good Judgment in the AI Era?"

HBR "AI-Generated Workslop Is Destroying Productivity"

HBR "Employees Are Relying on AI for Personal Support. That's Risky."

HBR "AI Doesn't Reduce Work — It Intensifies It"

BCG "AI at Work: Momentum Builds But Gaps Remain"

Microsoft Work Trend Index 2026

Upwork Research Institute

Deloitte "Human Skills Drive High-Performing Teams"

NBER "AI, Human Cognition and Knowledge Collapse"

Harvard Gazette "Deskilling Is Bad. This Is Worse."

Gallup "Rising AI Adoption Spurs Workforce Changes"

Ju & Aral "Collaborating with AI Agents"

If you're leading a team through an AI transformation right now, you're probably feeling it from both directions. The pressure from above to move faster — and the quiet uncertainty rippling through your team about what any of this actually means for them. Most of the conversation happening around you is about the technology: which tools, which platforms, which rollout timeline, which adoption metrics show you're moving fast enough.

What I want to talk about is the thing underneath all of that. The foundation that determines whether any of it actually works. After watching organizations try to do this, reading the research, and sitting with what I know about how people change — I'm convinced most organizations haven't built it yet.

The research points in a consistent direction. When managers visibly model AI use, employees show a 30-point lift in trust toward AI systems, according to the 2026 Microsoft Work Trend Index. That's not a training gap or a communication problem. The variable is who's leading — and how.

Meanwhile, teams are already moving without the official rollout. Between 78% and 86% of employees now use unapproved AI tools at work, regularly, because they've stopped waiting for their organizations to give them what they need. That ground-up movement is real and promising — and whether it becomes genuine transformation or fragments into isolated pockets of productivity depends almost entirely on what leaders do next.

BCG research put numbers to the gap: 76% of executives believed their employees were enthusiastic about AI adoption. The real number was 31%. Something is getting lost in translation — and it's getting lost at the manager layer.

The technology is not the constraint. The foundation is.

Before you keep reading: the framework below has four layers and more components than any one leader should try to change at once. Hold this as your lens — pick one trait from whatever layer lands for you, and give it a week. I wrote this for you so that you have a road map for what's actually important right now.

What Google Learned About Management

In 2008, Google launched Project Oxygen — an initiative to find out whether managers actually mattered. In a company built around brilliant engineers, the working assumption was close to the opposite: did management add real value, or just get in the way?

They analyzed 10,000 observations across thousands of employees — performance reviews, surveys, top-manager nominations — and ranked manager behaviors by their actual correlation with team outcomes. The results surprised everyone, starting with Google.

Technical skills ranked last. Dead last, out of eight behaviors.

The behavior that ranked first? Being a good coach.

Google's engineering culture had assumed the best managers were the ones who knew the most. What the data showed was that the best managers were the ones who asked good questions, invested in their people's development, communicated clearly, and listened — who showed up as human beings in relationship with other human beings, not as technical validators.

By 2018, Google revisited the list, expanding it from eight behaviors to ten. The two they added — collaborating across boundaries and being a strong decision-maker — turned out to be highly correlated with better team outcomes. The coaching still came first. The technical skills stayed last.

In 2026, a parallel shift is underway. The conversation about AI leadership has been dominated by questions of technical fluency — who understands the tools, who can speak the language. But Ethan Mollick, after running an experiment in which MBA students built functional startup prototypes in four days using AI, wrote something that stays with me: "The skills that are so often dismissed as soft turned out to be the hard ones." The students who succeeded weren't AI experts. They were people who already knew how to scope a problem, define what good looked like, and recognize when something was off. Management skills, it turns out, are AI skills.

The traits that make managers effective during AI transformation are not primarily technical. They are relational, emotional, and cognitive.

The Crisis Hiding in Plain Sight

Gallup's State of the Global Workplace 2026 released numbers that should give every executive pause. Manager engagement has fallen nine points since 2022 — down to 22% globally, the steepest sustained decline of any employee group Gallup tracks. In the United States, overall employee engagement has hit a 10-year low. And Gallup's decades of research on what drives team performance points to a single conclusion: managers account for 70% of the variance in team-level engagement.

The people who are supposed to be creating the conditions for AI adoption are running on empty.

This isn't just a morale problem. It's structural. HBR Study found Middle managers — the people closest to where AI transformation actually happens — feel less psychological safety than any other group in the organization. Less safe than the C-suite. Less safe than their own direct reports. They sit in the middle of every organizational decision, absorbing pressure from above and translating it downward, often without a coherent narrative to offer their teams.

The gap between what the executive floor believes and what the manager layer is actually living has never been more costly to ignore.

What You're Up Against

The fears employees are carrying right now are specific, not abstract. Trust in a direct manager is the strongest single predictor of whether an employee engages with organizational change — and the fear most people are navigating is concrete: Will I still have a job? Will I still be good at it? Will what I've spent years building still matter?

That fear doesn't announce itself. It shows up as compliance without contribution — people doing the minimum to appear AI-capable while protecting themselves from what they don't trust. Stanford psychologist Jamil Zaki, writing in HBR, noted that nearly a third of employees — and 44% of Gen Z workers specifically — admit to actively sabotaging their company's AI strategies. This is a rational response to feeling like the organization is not on your side.

At the Human+Tech conference in San Francisco, I heard John Hagel — whose book The Journey Beyond Fear traces what actually moves people through resistance — name the core tension: most people, looking at the future right now, see it as a threat. What overcomes fear isn't reassurance. It's the experience of having impact. People who can see their contribution mattering stop scanning for threats and start looking for possibilities.

Leaders who name the fear — who acknowledge it explicitly rather than manage around it — are doing something most leaders aren't: addressing the actual constraint.

And at a personal level, as a leader, the anxiety you haven't processed will travel downstream. That's not a leadership theory. That's just how rooms work. Burned-out managers produce disengaged teams, and disengaged teams do not build AI transformation.

What the Research Points To

The traits that matter for AI transformation leadership build in layers — each one enabling the next, inside out. Layer 1 is who you are. From there, how you learn, how you relate, and what you build follow.

Not all of this is equally in your hands. Some of what lives in the outer layers — team structure, autonomy, space to experiment — depends on what's above you. That's real. But the layers closest to the center are yours regardless of what the org chart says. And those are the ones that change the room.

Layer 1: Who You Are

This is the center — and the layer most organizations completely ignore.

Before what you do as a leader, there is who you are. The inner state you bring into the room. The stability underneath your decisions. The degree to which you've examined your own relationship to uncertainty, change, and not knowing.

Three things live here that matter enormously right now.

Passion is the most underrated. Not enthusiasm performed for an audience — genuine care about something. The work, the people, the possibility of what AI could open for your team. John Hagel's research on what he calls the passion of the explorer shows that leaders with this quality approach obstacles differently: their first instinct when they hit a wall is to find who else can help them get to a better answer faster, rather than to protect themselves. That orientation is contagious in both directions. Fearful leaders create fearful teams. Curious leaders create curious teams.

Growth mindset is the orientation that makes everything else possible. Carol Dweck's decades of research at Stanford and mindset book established that the belief our abilities can develop — rather than being fixed — fundamentally changes how people respond to challenge, failure, and uncertainty. In this moment, leaders with a growth orientation try things, share what they learned, and make it safe for everyone around them to do the same. Leaders without it tend to protect what they know and avoid what they don't.

Staying grounded under pressure is the competency almost no leadership curriculum addresses, and everyone needs it right now. The anxiety you haven't processed will travel downstream — your team feels it before you've said a word, and it shapes every decision you make when you're running on empty. The ability to be where the anxiety stops, rather than the channel through which it moves, is what keeps you curious, present, and open when your team needs you to be.

Layer 2: How You Learn

This is what your team is actually watching. Not what you say about AI — what you do with it. Whether you admit what you don't know. How you respond when something doesn't work.

The most important finding from the 2026 Microsoft Work Trend Index on what separates high AI performers from everyone else isn't which tools they use. It's that their managers are openly learning alongside them. When a leader shares what they tried and what didn't work, asks their team what they're discovering, and admits what they don't yet understand — that behavior creates permission at scale.

Ryan Vauk, who leads AI transformation at Google, described the shift at the Human+Tech conference: the managers making AI transformation real are moving from expert to catalyst, from analyst to storyteller, and — most importantly — from demonstrating perfection to demonstrating courage. In a world where no one yet knows the right path, the leader willing to not have all the answers is more valuable than the one who projects certainty they don't have.

Curiosity is the antidote to fear. The most important practice right now is using AI yourself — openly, imperfectly, in front of your team. That's what gives everyone around you permission to do the same.

Layer 3: How You Relate

This is where people decide whether it's safe to bring their whole selves. Empathy, trust, psychological safety, credibility, motivation, confidence — these are the relational conditions that determine whether your team contributes their real expertise or holds back.

Empathy is the one executives are most visibly abandoning right now — and the cost is measurable. A 2021 Catalyst survey found that 61% of employees with empathic managers reported actively innovating at work, compared to just 13% of employees whose managers were not empathic. The gap isn't personality. It's safety.

Psychological safety is the bedrock — and AI is creating new ways to erode it. When AI provides confident but incorrect information, teams don't just lose trust in the AI — they start losing trust in their own judgment. Harvard Business School professor Amy Edmondson calls this "trust ambiguity" — and it's one of the more insidious dynamics happening inside teams right now.

Motivation is the trait most connected to what Project Oxygen actually found. Coaching ranked first — and at its core, good coaching is about helping someone see why their work matters, not just requiring them to do it. In a moment when people are genuinely asking whether their contribution still has value — whether AI has made them less necessary — a leader who can help someone see what only they bring, what can't be replicated or prompted, is doing something that has always mattered and now matters enormously.

Confidence-building means helping your team separate skill-building from identity threat. ManpowerGroup's 2025 data found AI usage increased 13% while confidence in using those tools dropped 18% in the same period. Three-quarters of employees don't feel confident. What's happening underneath: people can't distinguish between "I'm not yet confident using this tool" and "I'm losing my professional edge."

Managers who can name that difference clearly — who can say "your expertise hasn't diminished, you're adding a new layer on top of it" — are doing something that directly determines whether people bring their judgment to AI work or just comply with the output.

Layer 4: What You Build

These are the structural and cultural choices — what the manager actively creates, protects, and measures for their team.

Collaboration over silos. Deloitte's 2026 research on AI and team structure found that high-trust teams use AI at 83% rates versus 63% for lower-trust teams, and cross-functional teams are 30% more likely to report significant efficiency and innovation gains. The manager is the structural orchestrator of whether people build on each other's work or protect their own lane. As Rosanna Durruthy observed at the conference: the teams that will matter most are the ones designed to need each other — where AI creates interdependence rather than isolation.

Autonomy. The ground-up AI movement happening right now is powered almost entirely by it. For the first time, people have tools they don't need to write a business case for — automation at their fingertips without asking permission. Teams given the autonomy to reorganize their work in ways that make sense for them are building genuine capability. Teams waiting for the approved rollout are falling behind.

Space to learn and innovate. Almost all AI investment goes to the technology itself — next to nothing goes to training and culture. The manager is the one who decides whether the time that learning actually requires gets protected or consumed by run-rate demands. John Hagel's model: small impact groups of three to fifteen people, given time and space to learn together and create new knowledge. The organizations that accelerate won't be the ones with the best tools — they'll be the ones with the most deliberate learning cultures, where experimentation is rewarded and failure is information rather than evidence of poor judgment.

This Isn't Just Change Management

Here is what I believe after watching organizations try to do this: most of them are approaching it backwards.

The conversation is almost entirely about what AI is going to do — to jobs, to workflows, to industries, to people's sense of who they are at work. And that conversation is real and worth having. The best leaders I'm watching right now have flipped the question. They're not primarily asking what AI is going to do to them or their teams. They're asking who they're going to be while it happens. How they're going to show up.

Change management matters. The communications plan, the rollout sequence, the champions structure — all of it is necessary. But it isn't sufficient. Change management puts the leader outside the process, architecting it toward a desired outcome. What this moment requires goes further than that. It's being in it with your team, not orchestrating them through it. It's showing up like a good human being — honest about what you don't know, genuinely curious about what's possible, willing to model the learning rather than perform the confidence.

And that learning has to be real. Leaders need to actually use AI — not delegate it, not observe it from a distance, not wait until they've mastered it before showing their team. The most powerful thing a leader can do right now is be visible in their own learning curve. Use a tool in front of your team. Share what surprised you. Say out loud what you tried that didn't work. That's not vulnerability for its own sake. That's the signal that tells everyone around you it's safe to do the same.

Leadership in most organizations was already flawed before AI arrived. AI doesn't create the gaps — the identity tied to having answers, the anxiety passed downward, the need to project confidence that isn't there. It just makes those gaps more visible and more costly.

The leader who doesn't do this doesn't just fall behind on adoption. They lose their credibility — the specific credibility that comes from being willing to do what they're asking their people to do. You're asking your team to grow publicly, to learn while uncomfortable, to stay curious when it's hard. If you won't go first, the ask doesn't land. It just becomes one more thing being done to people who are already overloaded.

Your people are watching every decision you make for signals about which side you're on. Whether it's safe to struggle. Whether it's safe to not know. Whether trying something that doesn't work will be held against them. Whether their expertise still matters. Whether you see them.

The most important thing you can do right now isn't to have the right AI strategy. It's to be the kind of leader who makes it possible for the people closest to the work to bring what they actually know to building it.

Pick one trait. Give it a week. Show your team what it looks like when the person who's supposed to have the answers is still learning.

That has never been a design question. It's always been a leadership one.

Sources

Google Project Oxygen (HBR, Dec 2013)

State of the Global Workplace 2026 (Gallup)

Managers Account for 70% of Variance in Employee Engagement (Gallup)

Middle Managers Feel the Least Psychological Safety at Work (HBR)

Work Trend Index Annual Report 2026 (Microsoft)

Empathetic Leadership Can Make or Break AI Adoption — Jamil Zaki, HBR April 2026

How to Foster Psychological Safety When AI Erodes Trust on Your Team — Seth & Edmondson, HBR Feb 2026

Bridging the AI Value Gap: Are Team Dynamics the Missing Link? (Deloitte, Feb 2026)

Employee AI Fears in 2026: What Actually Kills Adoption — ManpowerGroup data via People Managing People

The Journey Beyond Fear — John Hagel

Management as AI Superpower — Ethan Mollick, Substack, April 2026

Human+Tech Conference, San Francisco, May 2026 — John Hagel, Ryan Vauk, Rosanna Durruthy

My sister Lyn is a patient advocate. When my mother found a lump in her breast, the referral she got sent her to the wrong kind of doctor entirely — a specialist who had nothing to do with breast health.

Lyn didn't argue with the referral. She asked the receptionist a different question: "If your mother had a lump, who would you take her to?" She got a private call with three real names.

At the appointment with the right doctor, Lyn noticed something in my mother's chart that connected two things the doctors hadn't put together. She asked about my mother's other breast — the one nobody was there for. The doctor ordered an MRI.

Three spots in the other breast. Cancer, caught nearly a decade before it would have become something serious.

Lyn describes how she works: "It's things I have no idea are going to come up. I'm taking in my surroundings and what's going on in the conversation, and I come up with these questions without knowing what the question is."

That question — the one she didn't know she was going to ask — is exactly what AI transformation discovery is looking for. And almost never gets.

Documentation Is Not Discovery

Every organization has two versions of how work actually happens.

The first is the official version: the one explained in onboarding, described in meetings, drawn on process maps. Clean, sequential, logical. People have said it so many times it no longer sounds like a description. It sounds like policy.

The second version is what's actually happening. The workaround everyone uses without thinking about it. The judgment call made dozens of times a day that never got written down because it stopped feeling like a decision. The step that regularly stalls because someone specific isn't available, and the official process doesn't account for that.

A 2020 HBR study on digital transformation put the instruction plainly:

Look for what isn't quite working in your operation. Where do you regularly need workarounds or have to stop a process unexpectedly to fetch more information or involve another person?

Those workarounds are not inefficiencies to eliminate. They're signals. They're where the official version diverges from the real one — and where the real intelligence of the process lives.

When discovery surfaces the first version and not the second, the AI handles the documented steps correctly and quietly passes over the judgment, context, and implicit expertise that make those steps actually work. The output is technically correct and contextually incomplete. The team notices. They conclude the AI isn't capable enough. The real problem — thin discovery — never gets examined.

You Can't Prepare the Right Question

In the early 2000s, GE Healthcare engineer Doug Dietz spent years building an MRI machine he was proud of. Then he visited a hospital and watched a child walk toward it.

She was terrified. Sedation rates for pediatric MRI patients were above 90% — not because of the procedure, but because of the experience.

Dietz didn't ask patients how they'd improve MRI. He spent time with children. Heard their stories. Learned how they played and what made them feel brave. What he found: kids don't need calm. They need adventure.

The result was the Adventure Series: MRI suites became pirate ships and space stations. Sedation rates dropped to under 10%.

Harvard Business School research on thousands of natural conversations found something counterintuitive: the most powerful questions in a discovery session can't be prepared in advance.

Follow-up questions seem to have special power. They signal to your conversation partner that you are listening, care, and want to know more.

You can only follow up on what you just heard. Which means the right next question is only available to someone who walked in without a conclusion already waiting.

Steven Baert, former Chief People Officer at Novartis, named what most discovery interviews are doing instead:

"Previously [I focused on] listening to fix. 'You have a problem. I need a few points of data from you so I can solve the problem.' [But now] I'm practicing listening to learn."

Listening to fix already has the answer. It's collecting confirmation. Listening to learn is what surfaces the second version of the process — the real one.

What Discovery Actually Captures

The goal of a real discovery interview isn't a process map with more detail. It's a complete picture of how work actually lives inside the people doing it.

That means going further than steps and handoffs. It means understanding:

What people experience. Which parts of this process do they genuinely enjoy — the parts that feel like the real work they were hired to do? Which parts do they dread? Which feel necessary and which feel like overhead they've quietly resented for years?

How decisions actually get made. Not what the decision tree says. What does someone consider when something unexpected happens mid-task? What signals tell them to escalate versus handle it themselves? When do they trust their judgment and when do they check with someone else?

What people actually want to be doing. Not a leading question — a real one. What would this job look like if they could design it? What kind of work makes them feel good at the end of the day? What are they not doing right now that they wish they were?

Why they think each step exists. Some steps are genuinely necessary. Some are legacy. Some are there because someone who left three years ago needed them, and nobody has questioned it since. The person doing the work usually knows the difference. Nobody asked.

And then: all the follow-up questions in between. The ones that only become visible after the previous answer. That's where the second version lives.

Robin Dreeke spent his career as an FBI counterintelligence agent building trust quickly and getting people to share things they hadn't planned to share. In Warren Berger's The Book of Beautiful Questions, he described the prerequisite for this kind of listening:

"Am I genuinely interested in the other person? Am I able to put my ego aside and suspend all judgment? Am I prepared to truly listen, as opposed to just acting as if I am listening?"

Before you end every discovery conversation, ask the question that almost never gets asked: what didn't we talk about that would matter most? What are you not saying because you weren't sure it was relevant?

That's where Lyn lives. In the question that wasn't on the intake form.

Who Does This — and How Long It Takes

Most people weren't trained to do this kind of listening. That's not a failure — it's the norm. Most of us learned to come in with an agenda, run through it efficiently, and leave with clear notes. Following the conversation instead of the script is a different skill, and developing it takes practice and a specific kind of attention that most teams haven't been asked to build.

This is why discovery, at the start, needs someone from the outside.

An external person can see what your team has stopped seeing. They don't know which workarounds are "just how it is." They aren't carrying the organizational assumptions that make the second version invisible. They can ask the naive question — and mean it. They can notice the pause, follow the aside, and stay in the question long enough to find the one nobody knew to ask.

But the goal isn't permanent dependence on that outside person. It's the opposite.

The best engagements work like this: the consultant runs the first discovery rounds with your team alongside them, and makes every step visible as they go. The questions they ask, the patterns they're listening for, the moment they decide to follow a thread instead of moving on — all of it gets named explicitly. Your team watches, participates, and starts to build the muscle. After several rounds, they can do this without outside help.

That's the real deliverable: not just findings, but internal capability. The ability to keep doing this as the work evolves, as AI capabilities change, as new processes get built. This is also why AI adoption keeps stalling. Everyone can see what AI can theoretically do. Far fewer organizations have built the internal skill to know where it actually belongs — which requires understanding your own processes deeply enough to make that call. Discovery is the practice that builds that muscle. The upskilling isn't separate from the transformation. It's inside it.

As a rough guide: thorough discovery for one process — talking to multiple people, documenting what's stated and what's revealed — takes about a month. Design thinking and prototyping takes another two to four weeks. Building and testing takes roughly another month. One workflow, end to end: three months. And that doesn't include the pre-work that has to happen before discovery even starts: leadership alignment, a communications plan, change management, getting the champions team in place. That comes first. Discovery without organizational buy-in produces findings nobody acts on.

The timeline sounds long. It isn't, relative to what it replaces. A lot of AI transformations are failing right now — not because the tools don't work, but because they were built on the wrong picture of the work. The budget that goes into an implementation that doesn't stick dwarfs the investment in doing discovery right.

What Transforms vs. What Merely Improves

Jeremy Utley and Perry Klebahn, in Ideaflow, make an observation that applies equally here: "Ideaflow depends on inputs reaching a receptive mind." What comes out is determined by what goes in.

When discovery inputs are thin — official processes, prepared answers, the stated version — AI outputs are thin. You get faster documentation. Incrementally cheaper workflows. A slightly smoother version of the process you already had. That's useful. It's not transformation.

When discovery goes deep enough to find the real version — the judgment behind the steps, the expertise that's never been written down, the intelligence that makes the official process actually work — something different becomes possible.

Best Buy survived the Amazon threat not by optimizing its existing retail model, but by discovering what customers actually needed — something no process map would have revealed. When customers said they felt overwhelmed by technology, the obvious response was: improve the store. Better layout, clearer signage, faster checkout.

But that wasn't the insight. Customers didn't need a better retail experience at all. They needed someone to help them figure out what to buy, set it up at home, and fix it when it broke. They needed expertise, on demand, in their lives — not in a store. That discovery led to Geek Squad, in-home services, and an advisory model that turned Best Buy's physical presence into an advantage instead of a liability. They found the right problem by starting with the person, not the process.

That's the difference between transformation and digitization. And it gets decided in discovery.

The leaders who've moved fastest on AI transformation did the slowest, most careful discovery. They found the real version. That's what told them where to build.

You can't automate your way to the right problem. You have to find it first.

If you're wondering where to start: one workflow, one outside set of eyes, and a commitment to go deep enough to find the version that isn't on the map. From there, you build the capability internally — through champions who carry the change, through role design that protects the judgment AI shouldn't replace, through upskilling that gives your team the language and the practice to keep doing this on their own. Discovery is the first step. Everything else builds on what it found.

That's what the rest of this series is about.

Sources

"The Surprising Power of Questions" — Alison Wood Brooks & Leslie K. John, Harvard Business Review, May–June 2018

"The Art of Asking Smarter Questions" — Chevallier, Dalsace & Barsoux, Harvard Business Review, May–June 2024

"Discovery-Driven Digital Transformation" — Rita McGrath & Ryan McManus, Harvard Business Review, May–June 2020

The Book of Beautiful Questions — Warren Berger

Ideaflow: The Only Business Metric That Matters — Jeremy Utley & Perry Klebahn

GE Healthcare Adventure Series / Doug Dietz

On April 22, Meta sent its entire workforce a memo.

The company was installing software on employee computers that would capture mouse movements, keystrokes, clicks, and screenshots in real time. The purpose was to train AI agents to perform white-collar computer tasks autonomously. Employees could not opt out.

As Gizmodo put it:

Workers are "essentially being told they are training the systems that will replace them."

Meta's stated rationale was direct: "If we're building agents to help people complete everyday tasks using computers, our models need real examples of how people actually use them." That logic isn't wrong. The problem is what it reveals about how Meta sees the relationship — and specifically, what they decided to take from employees.

What Was Actually Taken

There's a question gaining traction in conversations about AI and the future of work: who owns the AI agents you build while you're employed somewhere?

A recent piece in Forbes by Nirit Cohen, she makes a compelling argument. When you build AI systems around how you think — systems that learn how you structure decisions, weigh tradeoffs, and navigate your specific role — those systems begin to encode something more than productivity. The author describes them as "a digital extension of how you operate professionally." Not just a work tool. A career asset that accumulates over time and could, in theory, travel with you across roles. The expertise you've built over years gets encoded into a system that starts to look less like company property and more like an extension of you.

Researchers are starting to call this cognitive sovereignty — the idea that as AI systems encode the patterns of how you think and work, the question of who controls that becomes genuinely unresolved. One academic paper cited in the Forbes piece frames it as "meaningful control over systems that hold your thinking and decision patterns." No legal framework has answered this cleanly yet. But the recognition that it's a real question is growing.

Meta didn't pause for that question. Their employees' professional patterns — the way they navigate their tools, structure their work, and make decisions in motion — get captured, encoded, and handed to systems those employees had no say in building, won't own, and won't share the upside of except for very small employee stock shares for most.

That isn't just a policy decision. It's a statement about what Meta believes their people are for.

Why This Breaks Something Specific

To understand what's really at stake here — not just morally but practically — it helps to know what trust violations actually do, and why some of them don't heal the way others do.

Research from Wharton professors Maurice Schweitzer, John Hershey, and Eric Bradlow — published in Organizational Behavior and Human Decision Processes — found that not all trust violations are the same.

The first kind is an ordinary failure. Someone misses a commitment, doesn't follow through on something important. That kind of trust can be rebuilt — consistent, trustworthy behavior over time does it, and a genuine apology can help speed things along.

The second kind is an integrity violation. That's when the failure removes someone's ability to say no, or involves deception. The Wharton research shows this kind causes "significant and enduring harm." Trust "recovered more slowly and less completely," even with genuine effort — and in some cases, it didn't fully recover at all.

What Meta did falls into the second category — not because anyone lied, but because the employees had no option to decline. When you remove someone's ability to say no, you've crossed into what the Wharton researchers would call an integrity violation.

Research published in Psychology Today in late 2025 adds a physiological dimension to this. When trust breaks seriously, the nervous system responds as if a threat is still present — people become hypervigilant, scanning for the next problem before it arrives. Real repair, when it happens, takes six to twelve months of consistent, genuine effort from the company. And the research is clear: the company doesn't get to decide when trust is restored. That belongs to the people who were hurt.

Meta defended the program. No apology, no opt-in alternative, no acknowledgment that this decision was a different category of choice. For Meta's employees, the repair clock isn't running, it hasn’t started and might not ever start.

The Cost Nobody Is Counting

I watched a version of this play out at VMware, at a much smaller scale.

We had a pay equity program designed for a genuinely good reason — when employees relocated across regions, salaries adjusted to cost-of-living benchmarks, and to make it work, people had to self-report their location every quarter. The purpose was fair compensation.

What people couldn't let go of was the repetition. And for many that understood cost of living indexes, at a human level they still struggled to see that if they were doing the same work, why would they make less moving a couple of states, as their output was the same regardless.

Teams that had been delivering remotely for years would get the form again — and with it, the unspoken message that they needed to be checked again. "Why are you asking me again?" That feeling of being rechecked, of not quite being trusted despite a track record, built over time. The people who felt it most were the ones who had performed consistently without oversight, and were still being asked to re-prove where they lived, quarter after quarter.

Another big thing came up here, I went to the VP of HR during the Great Resignation with a simple point: we're cutting the salaries of the people we most want to keep. Someone I had worked closely with — a Wharton recruit who was one of the highest performers on her team — asked to move states. Her salary ran through the index and dropped. I pushed back hard. She left anyway.

Over 15,000 people moved through that program. All we were tracking was where someone slept.

Here's the question I keep coming back to: is a nervous system in protection mode where people innovate? Is that the state where someone brings their most creative thinking to a hard problem, or has the courage to raise a concern when something is going sideways, or goes beyond the minimum of what they're asked to do?

It isn't. Every leader knows it isn't. The best companies invest enormously in employee engagement precisely because the research is clear — it's what drives quality of work, retention, the willingness to take on hard things, and the kind of creative contribution that actually moves companies forward. Meta just put all of that at risk. Not quietly, and not by accident.

And the fear your people are carrying isn't abstract. They're not worried about AI in general. They're thinking about specific questions that don't have answers yet: will I still have a job on the other side of this? At what salary — and will it be enough? Will the work even be something I care about anymore? What does this mean for my kids? Will I be able to maintain my quality of life or will I have to make sacrifices?

When that fear is already present — and you make an AI decision without explaining, without giving people a choice — you're not just creating a new trust problem. You're activating something that was already there.

The Version That Actually Works

Most people don't resist AI because they don't want to change. They want to be part of building something that matters — they want their expertise to count, their ideas to have a place, their contribution to be visible in what gets built. That instinct is an asset, and it's available to leaders who earn it.

What people resist is the feeling of being used rather than collaborated with.

The companies that move fastest on AI won't be the ones who extracted the most from their people the quickest. They'll be the ones whose people genuinely wanted to help build it — because the relationship was there, because someone had done the harder work of figuring out what it actually means to build with people rather than from them, and when the time came, people felt like partners rather than subjects.

Think about what that kind of contribution actually requires. People have to bring their real expertise — their honest judgment, the way they genuinely navigate their work, the insights that only come from years of doing something well. You cannot mandate that. You can require someone to use a tool. You cannot require them to bring their best thinking to it. And in any real AI transformation, their best thinking is exactly what you need.

When employees can see their expertise being used to build tools that make them more capable — when the upside is visible and shared, when what's being built is something they'd have wanted to help build anyway — they stop thinking about protection and start thinking about possibility. They'll help improve what gets built. They'll catch what the AI gets wrong. They'll bring ideas that couldn't have come from anywhere else. That's not a nice-to-have. That's the whole thing.

The difference between those two outcomes isn't the technology. It's whether people felt like they were a part of it.

This Isn't a Design Problem

The easy lesson here is process: build better opt-in frameworks, run cleaner pilots, get the communications right.

That's the design answer, and it misses what's underneath.

What happened at Meta — and what happens every time fear wins an AI decision — isn't a process failure first. It's a principles failure. The process reflects what a company actually believes about its relationship with its people. If the operating principle underneath is "we need this and we're going to take it," no rollout plan repairs what that signals.

The leaders navigating AI transformation well aren't doing it because they engineered better consent mechanisms. They're doing it because they genuinely understood what they were asking for — which is a great deal — and they showed up in a way that made it safe to give.

Your people are trying to figure out what their work life looks like on the other side of this. They've built their expertise and their sense of who they are around knowing how to do something well. They're not afraid of change. They're afraid of what this specific change means for them — and they're watching every decision you make to understand which side you're on.

How you show up in those decisions — what you explain, what you protect, what you ask rather than take — is what tells them.

You can mandate the tools. You cannot mandate the trust.

The fastest path is the one your people actually wanted to take.

Sources

"Mark Zuckerberg Just Sent a Shocking Message to Meta Employees" — The Street, April 22, 2026

"Promises and Lies: Restoring Violated Trust" — Schweitzer, Hershey & Bradlow, Wharton/Organizational Behavior and Human Decision Processes

"How Long Does It Really Take to Heal After Betrayal?" — Psychology Today, December 2025

"Can You Take Your AI Agents With You When You Leave A Job?" — Forbes, April 18, 2026

I had over $2,000 worth of executive coaching waiting for me — sessions already in progress, fully paid for by my employer. I wasn't going to waste them. I took every note. Documented every framework, every question that gave me insight..

I asked my coach if she’d want to explore building something like this together.

She said no. She was afraid it would cannibalize her business.

I understood. Fear does that. It makes us protect what we have instead of imagining what's possible.

But I still wanted it. So I turned the idea on myself instead. What if I documented how I think?

I started helping other people who'd been laid off — I'm good at seeing what someone has to offer when they can't see it themselves anymore. One person became a dozen. A dozen became a program. That program became the Phoenix Formula.

I'm telling you this because of something I read recently that put a lot of what I've been building into new words.

What the Sequoia Thesis Actually Says

Julien Bek at Sequoia published a piece called "Services: The New Software." The argument: the next category-defining company won't sell you a tool. It will sell you the outcome.

He uses two terms. A copilot is AI that works alongside you — making you faster, more capable, but keeping you in the seat. An autopilot sells the outcome directly. You don't hire someone to draft your NDA. You subscribe to "your contracts get done." As Bek puts it:

A copilot sells the tool. An autopilot sells the work.

To understand which work autopilots take first, he maps it along two axes: intelligence and judgment.

Intelligence is rule-based. Think of it as if this, then that — at any level of complexity. Coding a medical claim means matching a clinical note to one of 70,000 standardized ICD-10 codes. Filing a standard insurance form. Checking a contract against a compliance checklist. The rules are complex, but they are rules. AI can now follow them autonomously.

Judgment is different. It's experience, taste, and insight built over years of making calls and learning from the ones that went sideways. Knowing which decision changes everything. When the data says one thing but something in you knows better. How to read a room, a client, a moment. That's not a rule. It's something you build over time, often without realizing how much you've built.

One line stopped me:

Today's judgement will become tomorrow's intelligence.

That means the judgment you've built — everything you've learned to do that doesn't feel like a rule anymore because it's become second nature — is the most valuable raw material in this moment. More valuable, not less.

This Was Already Predicted

What makes Bek's thesis land so hard isn't that it's new. It's that it confirms what was already being documented — before we had the tools to actually do it.

In Digital Hesitation, Thomas Lah and J.B. Wood argued that B2B buyers had already stopped thinking about what tools could do and started thinking about outcomes — what will actually get done, and by when. Their XaaS thesis said the entire customer relationship was moving from capability to result.

When you sell an outcome, the buyer expects it on a timeline. Think about Amazon two-day shipping — you don't wonder if your package is coming, you just know it is. Time-to-value stops being a metric and becomes the whole product. More than that: once the outcome is guaranteed, how fast you deliver it becomes the competitive edge. The professionals who've figured out which pieces of their judgment to automate will get there faster.

The catch right now: a lot of that value never gets realized. Companies are paying for Claude Enterprise, ChatGPT, Microsoft Copilot. The tools are live. The utilization isn't. The upleveling of the team isn't there. People have access and no real framework for using it — and nobody is on the hook for the outcome yet.

That gap is real. And it's also an opportunity for anyone willing to close it.

The first autopilots aren't hypothetical. They're funded, operating, and already taking the work.

The Map and the People Inside It

Bek's framework for deciding where autopilots win first is one of the most useful things in the piece: start where work is already outsourced. His map covers insurance brokerage, healthcare billing, accounting, legal transactional work, recruiting — industries where someone is already paying for a result, not a person. Replacing that vendor with an AI-native service is a vendor swap, not a reorganization. That's the wedge — the lowest-friction entry point.

One thing I'd add: if you're looking at your own operations and considering outsourcing something today, I'd stop. When you hand a process to someone outside your walls, you lose visibility into how it actually works. And you cannot put something on autopilot that you no longer understand. Instead of outsourcing, invest in AI consultants who build the capability alongside you — so when autopilots arrive for that work, you're the one deploying them. There's a real difference between a consultant who builds with you and one who builds for you. The second creates the same dependency you were trying to avoid.

The people inside those industries can feel the shift. The jobs are changing before the press releases say so.

I had strong offers coming out of VMware — that was my experience. But what I hear from the people I work with now tells a different story. Professionals who earned $180,000 interviewing for $90,000. Senior people taking roles two levels below where they were.

The reason isn't complicated. Companies are making room for AI. They need the cost savings. And given the state of the market right now, they can get highly experienced people for a fraction of what those roles used to pay.

A lot of these people already made sacrifices. They took jobs that weren't their passion — work that wasn't their dream, but they were good at it and they built a life around it. Now they're finding that even those jobs, the ones they showed up for every day and got genuinely good at, aren't worth what they used to be. And on top of that, they're being handed AI tools and told to use them — to automate the very tasks that made up their job descriptions.

There's something else worth naming. If your judgment trains a model — the patterns you've built over decades, the hundreds of thousands of dollars you've put into education and professional development — who gets compensated for that?

Tim Rayner has been thinking about this out loud. He calls it cognitive sovereignty — the right to control, protect, and license your professional expertise. Right now, teams are being handed AI mandates with no additional pay, no equity in what gets built from their expertise. When your knowledge becomes someone else's infrastructure, that's not deployment. That's extraction. We don't have an answer yet. But it's the right question to be asking.

What to Delegate and What to Keep

Some of your judgment can be automated — and doing it right makes you more powerful, not less.

There are two kinds of judgment-adjacent work.

The first is thinking that's become routine. The patterns you run through every time. The diagnosis you've made so often you don't think about it anymore. Your wisdom made systematic. That part can be documented, handed to AI, and scaled.

The second is the real-time read. What's actually happening with this specific person, in this specific conversation, right now. What they're not saying. The thing you catch before they do. That's the judgment you keep — and when you automate the first kind, you have far more space for the second.

There's a third kind that's harder to see and worth protecting: the experience that lets you look around corners. You've been in enough rooms, made enough calls, seen enough cycles play out — that you can look at a situation and have a read on what's coming. Not because you have data, but because you've seen this before. When a product team has lived through five launches, they know which customer complaints are noise and which ones are the signal of something bigger. When a sales leader has seen a market turn, they can feel the next one coming before the numbers show it. That's not intelligence. That's accumulated pattern recognition — and it's yours.

Paul Daugherty calls the space where this plays out the "missing middle" — where humans and AI extend each other's capabilities rather than one replacing the other. The way to find your missing middle is to decompose your role the way you'd build a team. If I had unlimited budget and could bring in anyone, who would I want for each piece of this work? That question separates your intelligence tasks — the complex but rule-based work AI can take — from your judgment tasks. The judgment tasks are the work you actually want to do. When you automate the intelligence layer, you free up capacity for far more of them.

How I Built the Phoenix Formula

I tested this in the domain that everyone assumes is permanently human.

I'd spent 15 years investing in my own career development — executive coaching, leadership programs, all of it. And I'd worked with hundreds of professionals through career transitions. When I started my own job search after VMware, the gaps in how people were being supported were impossible to miss. The tools existed. The frameworks didn't.

I wanted to see if I could automate the pieces of my judgment that were consuming my time, so coaching could be the highest and best use of the hours we actually had together. I built a main AI coaching model that clients could interact with between our sessions — it knew their situation, their progress, what they'd been working through. By the time we got on a call, I already had AI-surfaced insights about where they were stuck. What would have taken the first thirty minutes to uncover, I could get to in five. Ten hours of coaching work, compressed into one.

But before I could build any of it, I had to solve a measurement problem. Feelings aren't proof. Career coaching produces a feeling — people feel better, more confident, more clear. I needed to know whether the inputs that actually lead to getting hired were moving. That became the foundation.

Here's what I did.

I voice-memoed my root cause analysis — a real account of how I think about transformation, not a framework. Then I fed that to AI, asked it to act as a measurement expert using Douglas Hubbard's framework from How to Measure Anything, and had it design my assessment questions from my own thinking. Then I ran my call transcripts through AI to surface the patterns in how I was actually making decisions — the judgment showing up in real time, not just what I said I believed.

Then I built specialized agents — one for each role in the job search process. Ten thousand hours makes you a master in one thing. A great resume writer is not a great interviewer. A great LinkedIn strategist is not a great mindset coach. If you hired the best possible person for every single aspect of a job search, you'd have a team no one person could be. The Phoenix Formula gives every participant access to that team.

What I found: I didn't know the full architecture of my own thinking until AI surfaced it. The judgment was already there — I just couldn't see all of it from inside my own head. AI gave it language, structure, and the ability to reach people I never could have reached one-on-one.

The pre/post results, measured before a single participant had gotten a job, showed a 300% improvement in LinkedIn proficiency, 150% improvement in articulating unique value, 75% reduction in burnout, 80% reduction in sleep disruption. I measured inputs, not outputs. If the inputs move, the outcome is a matter of time.

As AI takes on more intelligence work, judgment gets pushed to a higher level. For anyone who's built real expertise, that's good news.

The work that becomes ours is more creative, more consequential, more human. Design thinking lives here — not optimizing what already exists, but looking around corners. Understanding what people actually need before they can say it. Designing what comes next rather than reacting to what already happened.

The ones who stay valuable are the ones who keep upleveling — using AI to compress the intelligence work so they have more space for the judgment above it.

Document your judgment. Voice-memo your thinking before you open a prompt. Ask AI for insights based on how you think — to surface what you already know, not just to generate output.

What you've built over your career is the most valuable thing in this equation. The question is whether you've made it visible enough to use it.

More capable. Not more dependent.

Sources

Julien Bek, "Services: The New Software" (Sequoia Capital)

Thomas Lah and J.B. Wood, Digital Hesitation

Paul Daugherty, Human + Machine

Tim Rayner, "We Need to Talk About Cognitive Sovereignty"

The AI mandate came down. The tools are there. The training happened.

No one handed you a playbook for the part that comes next.

Maybe your company issued the mandate and understand the technology — but not the people side of what you're asking your team to do. Maybe you're trying to support a team at five different levels of readiness with no shared framework for any of it. Maybe you're the one in charge of the initiative, the pressure to deliver is real, and the path isn't as clear as it looked when you said yes.

Why the Tools Aren't Enough

Daily AI usage has doubled since 2024. About 90% of workers are using large language models in some form. And yet only 2 to 4% of executives are seeing real, transformative organizational change. Ninety-five percent of AI pilots fail to scale.

McKinsey's research shows that workflow redesign — not tool deployment — is the strongest predictor of financial returns from AI. The organizations getting results are doing something both technologically and structurally different. They're not deploying tools into existing work — they're redesigning how work flows around what AI can actually carry.

And what AI can carry keeps changing. The capability ceiling is rising faster than most organizations can redesign around. Part of building the muscle is knowing what AI is ready to carry right now, in your specific environment — and revisiting that as the capability shifts.

The Fear Underneath the Mandate

There's something underneath the mandate that doesn't make it onto the slide deck.

The people on your team aren't afraid of AI as a tool. They're afraid of what work gives them — the income, the identity, the place in the organization they've spent years building. The rules they've been working within are changing, and no one has named the new ones yet. That fear is rational. And leaders are the one holding it for them — while simultaneously carrying the technology pressure, the results pressure, and the pace of a field that rewrites itself every few months.

Maximizing output in that environment doesn't produce transformation. It produces burnout. What people and organizations actually need is optimal performance — the right work allocated to AI, the right work kept with people, and a pace that's sustainable enough to actually move the needle.

That's the design problem. And it's the one most organizations haven't started yet.

Three Lenses

I'm Christine Reichenbach. I spent close to five years at VMware — leading Future of Work strategy for a global workforce of 37,000, then serving as Chief of Staff to the VP of HR through one of the largest tech acquisitions in history. Before that, process transformation at scale. I'm Stanford-certified in design thinking, Six Sigma trained, and for the past two years I've been building AI systems from the ground up — testing, failing, rebuilding, and teaching professionals how to lead AI transformation in their own organizations.

The organizations doing this well are treating AI as a design problem, not a deployment problem. Both the technology and the structure of work are changing. What stays constant is the need to design deliberately.

I look at every AI question through three lenses.

Design thinking: what do your customers and employees actually need? Start there. Design backward from that, with AI as one of the tools you use to deliver it. Organizations that skip this step automate the existing process and wonder why nothing meaningfully changed.

Six Sigma: how does work actually flow? You can't automate what you haven't mapped and measured. The process has to be understood before it's handed to AI. Most failed implementations skipped this work entirely.

Organizational design: who should do what — and at what pace? Not just "can AI do this" but should it? What judgment stays with people? What does AI carry? How do you manage the complexity of a field moving faster than most teams can redesign around, without burning out the people doing the moving? These decisions have to be made deliberately, or defaults fill the gap — and default settings don't optimize for people.

AI Organizational Design is what you get when all three work together. That's the piece that moves the needle — that takes AI from a tool your team uses to a capability that actually produces the outcomes and impact you're after.

One thing I've seen consistently: when people step outside their own role context — when they're solving a shared problem instead of defending their current one — the constraints drop and they discover what they're actually capable of. That's by design too.

What Intentional Design Is

Intentional Design is where I apply this publicly.

Every week: one piece of AI news or research worth your attention, run through that framework, translated into what it actually means for your business. Some weeks one article. Some weeks a few things that connect. I follow the content, not a predetermined format.

This is for business leaders and department heads who want to lead AI transformation well — not just deploy tools and hope, but redesign work deliberately, build real capability inside their teams, and make more impact with the time they have.

Beyond Business

Intentional Design isn't just a business framework for me — it's how I lead my life.

Every day I protect time that belongs to me. No output, nothing scheduled. Maybe I finally read the article that's been open in my browser all week. Maybe I need a meditation because I'm foggy, or something to get my brain back on track before the next thing. Maybe it's a random walk with a friend. I decide in the moment, based on what I actually need. Making that space is the practice.

I have three kids four and under. I live in Mesa, Arizona. Staying active and mindfulness aren't things I fit in around work — they're part of how I stay clear enough to do the work.

I mentor a teenager in foster care and helped her start her own flower business. And I'm working with other consultants to bring Intentional Design into the foster care system: better pay for workers, better outcomes for the kids we're responsible for, a system that actually moves the needle for the people who need it most.

Everything I build — this newsletter, the consulting work, the courses — comes back to one thing: helping people have more fulfilling lives. Not just more productive ones. More intentional, more impactful, more theirs.

You don't have to have your AI strategy figured out to be here. You just have to be willing to think about it differently.

I'm glad you're here.

More capable. Not more dependent.

Sources

Stanford HAI 2026 AI Index Report

McKinsey: The State of AI

Nobl.io: The State of Work Redesign, March 2026