Coldcode Si Labs Library

Replaced

Courage Garikai · Harare, Zimbabwe

REPLACED

A GUIDE TO THE AI-ERA WORKPLACE REPLACEDYou're Not Going to Be Replaced by AI.

But You Might Be Replaced by Someone Who Uses It.

COURAGE GARIKAI A GUIDE TO THE AI-ERA WORKPLACE REPLACED You're Not Going to Be Replaced by AI.

But You Might Be Replaced by Someone Who Uses It.

COURAGE GARIKAI First edition, 2026.

© 2026 Courage Garikai. All rights reserved.

Paper ISBN: 979-8-9912-0004-2

eBook ISBN: 979-8-9912-0005-9

REPLACED 2

This book is intended to provide helpful information on the subject matter covered. It is sold with the understanding that the publisher and author are not engaged in rendering legal, investment, or professional services. The research cited in this book is summarized in good faith; readers are encouraged to consult original sources and to adapt any recommendations to their individual circumstances.

Courage Garikai Books REPLACED 3

REPLACED

REPLACED You’re Not Going to Be Replaced by AI. But You Might Be Replaced by Someone Who Uses It.

Courage Garikai For every worker staring at a screen, wondering what happens next.

For everyone who has felt the floor shift under their career.

And for the people who will decide not to be replaced.

Contents PART I: THE SHIFT 1. The Sentence That Changes Everything 2. The Data: What’s Actually Happening to Work 3. The Productivity Multiplier — And Why It’s REPLACED 4

Not Evenly Distributed 4. The Myth of the AI Apocalypse (and the Real Threat Right in Front of You)

PART II: THE DIVIDE 5. The New Productivity Gap 6. Who Wins, Who Loses, and Why 7. The Skills AI Can’t Touch — And the Skills It Makes More Valuable 8. The Five Traps That Get Good People Fired First 9.

The Hidden Cost of Outsourcing Your Thinking PART III: THE PLAYBOOK 10. The AUGMENT Framework — Six Moves That Separate Thrivers from Survivors 11. Your First 30 Days: From AI-Curious to AI-Competent 12. Rewiring Your Daily Work: Before, With, and After AI 13. Building Your Human Moat: The Things AI Cannot Give Your Competitor 14. Making Yourself Visible in an AI-Augmented Organization 15.

The Side Hustle, the Portfolio, and the Safety Net PART IV: THE LONG GAME 16. Becoming the Person Who Directs the Tool 17. Leading Teams in the Augmented Age 18. Staying Human When the Machines Get Better Every Year Afterword: The Choice in Front of You Appendix:

The Evidence — Studies Cited in This Book Notes Acknowledgments REPLACED 5

PART I: THE SHIFT

PART I: THE SHIFT Chapter 1: The Sentence That Changes Everything There is a sentence circulating through offices, Slack channels, LinkedIn posts, conference keynotes, and dinner parties in every major city on earth. You have probably seen it. You may have nodded at it. You may have forwarded it. It goes like this:

“AI is not going to replace you. A person using AI will.”

Most people read that sentence, feel a brief jolt of recognition, share it, and then go back to doing their job exactly the way they did it yesterday.

That is the single most dangerous response you could possibly have.

Because this is not a motivational quote. It is not a meme. It is a description of a labor market that is already restructuring itself around you — faster than most people realize, and with consequences that will shape the next decade of your working life.

Let me say it clearly, so there is no ambiguity:

REPLACED 6

The threat is not a robot coming for your job. The threat is not a sentient algorithm deciding you are redundant. The threat is the colleague two desks over who started using AI three months ago and now does in two hours what takes you two days. The threat is the freelancer on another continent who underbids you by 60% because AI lets them deliver three times the volume. The threat is the recent graduate who entered your industry last year already fluent with AI tools and is producing the output of someone with five years of experience. The threat is the founder who used to need a team of ten and now runs a profitable business with three people and a stack of AI tools.

They are not coming to replace you with a machine.

They are replacing you with a person who has learned to multiply themselves with a machine.

And the worst part — the part almost no one is talking about honestly — is that this is not some hypothetical future scenario. It is already happening.

The data is in. The studies have been run. The productivity gains are being measured. The wage premium for AI fluency is already real. The layoffs have started, and they are not targeting the people you think.

This book is about what you actually do about it.

It is not a book about how AI works technically. It is not a book about prompt engineering tricks that will be obsolete in six months. It is not a book predicting a REPLACED 7

science-fiction future. It is a book about right now:

what is happening to work, who is winning and who is losing, what skills will actually protect you, and what concrete steps you can take starting this week to end up on the right side of the fastest redistribution of productive power in a generation.

Here is what I want you to understand before we go any further: the sentence at the top of this chapter is not a warning about AI. It is a warning about complacency. The technology is neutral. It will be used. The question is whether you will be the one using it, or the one it is used against.

You have a choice. You always have. But the window for making that choice proactively — for learning the tools, rewiring your workflow, building the skills AI cannot replace, and positioning yourself as someone who multiplies value rather than someone whose tasks can be multiplied by someone else — is narrower than you think.

Let me show you the evidence. Then let me show you the playbook.

REPLACED 8

Chapter 2: The Data: What’s Actually Happening to Work

Chapter 2: The Data: What’s Actually Happening to Work If you listen to the loudest voices in the AI conversation, you hear two contradictory stories.

The first is the utopian story: AI will produce infinite abundance, cure disease, free humanity from drudgery, create a world of leisure and creativity.

The second is the apocalyptic story: AI will take every job, create mass unemployment, concentrate all wealth in the hands of a few tech companies, and leave most people economically irrelevant.

Both stories are wrong. Both are distractions from what is actually happening in the data, which is more interesting, more nuanced, and — for the individual worker — more actionable than either extreme.

Let me give you the numbers. They come from peer-reviewed studies, field experiments, and labor market analyses conducted between 2023 and 2026 at institutions including MIT, Stanford, Harvard Business School, the National Bureau of Economic Research, GitHub, Microsoft, PwC, the World Economic Forum, the OECD, the Yale Budget Lab, and the Brookings Institution. I have cited every source in the appendix. I am not giving you futurism. I am giving you measurements.

REPLACED 9

What the field experiments show

WHAT THE FIELD EXPERIMENTS SHOW Study 1: 5,179 customer support agents.

Researchers at Stanford and MIT (Brynjolfsson, Li, and Raymond, 2023/2025) studied a staggered rollout of a generative AI assistant among customer support agents at a Fortune 500 software company. The tool suggested responses in real time, drawing on the company’s knowledge base and successful past resolutions. The result: productivity, measured by issues resolved per hour, rose by an average of 14%.

For novice agents with less than two months on the job, productivity rose by 34%. The least-experienced workers improved both their speed and the quality of their output, essentially absorbing the tacit knowledge of high-performing tenured agents through AI suggestions. For experienced workers, the gains were smaller but still positive.

Study 2: 453 college-educated professionals.

Noy and Zhang at MIT ran a preregistered online experiment published in Science (July 2023). They gave 453 educated professionals — writers, analysts, managers, marketers — incentivized writing tasks drawn from their actual occupations. Half were randomly given access to ChatGPT; the other half were not. The AI group completed their tasks 40%

faster on average, and their output, graded by independent evaluators, scored 18% higher in quality.

REPLACED 10

Study 3: 756 BCG consultants. Dell’Acqua and colleagues at Harvard Business School and Boston Consulting Group ran a controlled experiment with 756 BCG consultants across 18 realistic consulting tasks. On tasks within the AI’s capability range, consultants using GPT-4 completed 12.2% more tasks, finished 25.1% faster, and produced significantly higher-quality solutions than those working without AI. But — and this is a critical “but” — on one complex task deliberately chosen to sit outside the AI’s reliable range, consultants using AI were 19 percentage points less likely to produce correct solutions. The tool failed persuasively. It did not say “I don’t know.” It generated confident, plausible-sounding wrong answers that even smart people accepted.

Study 4: 4,867 software developers. Peng, Cui, and colleagues ran three field experiments with developers at GitHub and elsewhere using an AI coding assistant (the precursor to GitHub Copilot).

Developers using the assistant completed 26% more tasks than the control group. The gains were largest for less experienced developers and for developers who adopted the tool most deeply. A more recent METR replication found experienced developers initially 19% slower — but noted the result did not hold up by late 2025, as developers grew more fluent with the tool and the model itself improved.

Study 5: 88% adoption, only 5%

transformation. The EY Work Reimagined Survey REPLACED 11

(2025), covering thousands of workers across 25

countries, found that 88% of employees now use AI at work. But the vast majority use it only for basic tasks — search, summarization, email drafting. Only 5% use it in advanced, workflow-transforming ways that change how they actually produce output. The gap between casual use and mastery is where the competitive advantage lives.

Study 6: The wage premium is already here.

PwC’s Global AI Jobs Barometer (2025) found that workers with AI-related skills are earning up to a 56%

wage premium over otherwise-similar workers without those skills — and that this premium has roughly doubled in a single year. Sectors most exposed to AI are seeing both faster employment growth and faster productivity growth. The Dallas Federal Reserve found that even in heavily AI-exposed industries, experienced workers with AI skills are seeing wage growth, not decline.

Study 7: The jobs apocalypse that hasn’t arrived. A joint analysis by the Yale Budget Lab and the Brookings Institution (2025) found no significant overall impact of AI on aggregate employment since the release of ChatGPT in late 2022. Challenger, Gray & Christmas tracked 17,375 job cuts explicitly attributed to AI through September 2025 — real pain for the people affected, but a small fraction of the nearly one million total layoffs tracked in that period.

The World Economic Forum projects 92 million roles REPLACED 12

What the numbers actually mean

displaced and 170 million new roles created by 2030, for a net positive of 78 million jobs globally.

WHAT THE NUMBERS ACTUALLY MEAN Let me summarize what this body of evidence tells us, plainly:

AI makes individual workers measurably more productive on a wide range of cognitive tasks. The effect is real, it’s been replicated across multiple settings, and it clusters between 14% and 40% depending on the task and the worker’s starting skill level.

The gains are largest for the people who were previously behind. Novices benefit more than experts. Workers with weak writing skills gain more than strong writers. Developers with less experience gain more than veterans. AI is a skill compressor that transfers tacit knowledge from high performers to anyone who can use the tool well.

AI fails in predictable ways. On tasks outside its reliable range, it produces confidently wrong answers. People who trust it blindly get burned. People who verify, direct, and curate its output capture the gains without the failures.

1.

2.

3.

REPLACED 13

The macroeconomic effects are slow.

Despite three years of rapid AI progress, economy-wide productivity gains and job displacement are modest. This is consistent with every previous major technology — it takes years for organizations to redesign workflows, retrain workers, and capture productivity gains.

The steam engine, electricity, and the personal computer all followed the same adoption curve.

But the microeconomic effects — at the level of individual workers competing for individual jobs — are already sharp. The labor market doesn’t wait for macroeconomic transformation. If two candidates apply for the same role and one can do 1.3x the output of the other because they work with AI, the employer does not need to wait for a societal shift. They hire the augmented worker.

A wage gap is opening fast. The PwC finding of a 56% wage premium for AI-skilled workers is not a prediction about the future. It is a measurement of the present. Within individual companies, within individual professions, people who can multiply their output with AI are already being paid more, promoted faster, and retained preferentially.

4.

5.

6.

REPLACED 14

Most people are not serious about adopting AI yet. The EY finding that 88% of workers have dabbled but only 5% have truly integrated AI into their workflows is the most important number in this chapter. It means the window is still open. The competitive advantage goes to the people who move from casual use to deep integration before everyone else does.

Let me be blunt about what this adds up to.

AI will not replace you in the next year. It will not replace most people in the next five years. But a person who uses AI well will replace someone who doesn’t, because they will produce more, faster, cheaper, and often better — and employers, clients, and markets always reward that combination.

The first step to protecting yourself is to stop thinking of AI as a threat or a savior and start thinking of it as a new baseline skill — like using email, like using a search engine, like using spreadsheets a generation ago. There was a time when you could opt out of learning email and still have a career. That time ended. The same thing is happening with AI. The question is not whether you will use it. The question is whether you will use it well enough that you are the one doing the replacing, not the one being replaced.

7.

REPLACED 15

Chapter 3: The Productivity Multiplier — And Why It’s Not Evenly Distributed

Chapter 3: The Productivity Multiplier — And Why It’s Not Evenly Distributed To understand what’s happening to the labor market, you have to understand one concept: the productivity multiplier.

A productivity multiplier is anything that lets one person produce more output per unit of time than they could before. The wheel was a multiplier. Writing was a multiplier. The printing press, the steam engine, the assembly line, the telephone, the spreadsheet, the internet — every one of these was a multiplier. They let a person do more with less.

What makes AI different is the breadth of the multiplier and the speed with which it’s arriving.

Previous multipliers tended to multiply one kind of capability. The assembly line multiplied manual manufacturing output. The spreadsheet multiplied numerical calculation. The telephone multiplied communication distance. AI multiplies across the entire range of cognitive labor — writing, analysis, research, coding, summarization, translation, image creation, planning, customer communication, drafting, pattern recognition, basic decision support. It is the first multiplier in history that applies to the work of REPLACED 16

Why multipliers redistribute advantage

knowledge workers across almost every function simultaneously.

And it arrived in the space of about three years.

WHY MULTIPLIERS REDISTRIBUTE ADVANTAGE Here is the thing about productivity multipliers that most people miss: they do not benefit everyone equally. They benefit most the people who are already positioned to use them well. And they punish most the people who are best at doing things the old way.

When the spreadsheet arrived in the late 1970s and early 1980s, it did not benefit every accountant equally. The accountants who learned VisiCalc and Lotus 1-2-3 and eventually Excel multiplied their productivity dramatically. They could build models in hours that previously took weeks. They could do scenario analysis that was simply not feasible by hand.

Those accountants got promoted, got raises, started consulting practices, and captured the surplus. The accountants who resisted spreadsheets, who insisted that hand-calculated ledgers were more reliable, who refused to learn the new tool — they did not vanish overnight. But over a decade, their work was absorbed by smaller numbers of people using spreadsheets.

Their relative value fell. Their careers stalled.

The same story played out with word processing replacing the typing pool, with CAD replacing manual REPLACED 17

The unequal distribution of the AI dividend

drafting, with digital photography replacing film processing, with email replacing the mailroom. In every case:

The tool created real productivity gains.

The gains did not arrive everywhere at once.

The people who adopted early captured outsized rewards.

The people who resisted lost ground slowly, then all at once.

The net effect on total jobs was eventually positive — but the transition destroyed individual careers that did not adapt.

AI is following exactly this pattern. But because the multiplier applies across so many more functions simultaneously, the compression is faster.

THE UNEQUAL DISTRIBUTION OF THE AI DIVIDEND Let me walk you through why the gains are so uneven.

The novice boost. As the Brynjolfsson customer support study showed, the biggest productivity gains accrue to the least experienced workers. Why?

Because AI can transfer tacit knowledge. The high-

performing agents had internalized, over years, what to say, how to de-escalate, what solutions to offer for which problems. The AI system could encode that knowledge and offer it in real time to a brand-new •

REPLACED 18

agent. This is genuinely good news for new workers:

the learning curve is shorter. But it is bad news for workers whose only competitive advantage was having five years of experience doing routine tasks that AI now makes accessible to a rookie.

The expert augmentation. At the other end of the spectrum, genuine experts also benefit — but differently. An expert using AI does not get 34%

better; they get the ability to tackle larger, more ambitious problems. A senior lawyer using AI can review more documents in less time, freeing hours for higher-value strategic counsel. A senior developer using Copilot can handle more of the routine code, spending more time on architecture. The expert’s judgment, combined with AI’s speed, is a formidable combination. The PwC finding of a 56% wage premium for AI skills concentrates disproportionately here —

senior people with domain expertise who augment with AI can produce output that previously required a team.

The squeeze in the middle. The group under the most pressure is what you might call the competent routine middle — workers who are solid performers doing structured cognitive tasks (drafting standard documents, basic analysis, routine coding, standard customer service, entry-level design and marketing work) but who are not deep experts and are not novices. Their work was previously a barrier to entry:

you needed a certain amount of training and REPLACED 19

experience to do it competently. AI lowers that barrier.

A junior worker with AI can produce output that previously required three to five years of experience.

An expert with AI can absorb more of their routine work. The middle gets compressed from both sides.

The adoption gap within roles. Perhaps the most underappreciated finding: within the same job title, at the same company, in the same department, there are people who have reinvented how they work using AI and people who have not. These workers are ostensibly doing the same job. But one group is producing 1.3-2x the output of the other. This is unstable. Managers notice. Budgets get reallocated.

Promotions accrue to the augmented. Layoffs, when they come, fall first on the un-augmented.

The 5% problem. Remember the EY finding: 88%

of workers use AI, but only 5% use it in ways that actually transform how they work. Most people use AI to write a first draft of an email, summarize a long document, or answer a factual question. That is the equivalent of using a computer as a fancy typewriter.

The transformational gains come when you redesign workflows around AI: when you delegate entire categories of cognitive work to it, when you chain tools together, when you build systems, when you use AI to iterate faster, test more ideas, make better decisions. That level of fluency is rare, and it is where the competitive moat lives right now.

REPLACED 20

What this means for you

WHAT THIS MEANS FOR YOU The productivity multiplier is real. It is here. It is not evenly distributed.

If you are a novice: AI is the greatest opportunity you have ever had. You can close the experience gap faster than any previous generation could. But you must actually learn the tools deeply — not just use them casually — and you must build the underlying judgment that prevents AI from leading you astray.

If you are an expert: AI is an amplifier. It lets you do more of what only you can do. It can multiply your impact, your income, and your reach. But only if you are willing to change how you work.

If you are in the middle: you are in the danger zone.

Your routine competence is no longer enough. You must either move toward expert judgment and human differentiation, or move toward mastery of AI-

augmented workflow, or ideally both. Standing still is the riskiest option.

The rest of this book is about how to move deliberately, starting this week, toward the side of the divide where you are the one using the multiplier —

not the one being multiplied against.

REPLACED 21

Chapter 4: The Myth of the AI Apocalypse (and the Real Threat Right in Front of You)

Chapter 4: The Myth of the AI Apocalypse (and the Real Threat Right in Front of You)

It is tempting, when confronted with a technology this powerful, to reach for one of two extreme narratives.

Let me dispose of both, because both of them will prevent you from taking the right action.

NARRATIVE ONE: THE JOB APOCALYPSE The first narrative says AI is going to take almost every job, creating mass unemployment, universal basic income will become mandatory, and most people will be economically irrelevant. This story gets clicks, sells books, and drives conference keynotes. It is also, based on every piece of credible data we have, wrong as a description of the next decade.

The historical record on technology and employment is remarkably consistent. Every transformative technology — the steam engine, electrification, the automobile, the personal computer, the internet — caused widespread predictions of mass unemployment. None of them produced it. What they produced was a restructuring of work: some jobs vanished, others were created, the composition of the workforce changed, and the transition was often brutal for specific groups of workers. But total REPLACED 22

Narrative Two: The Productivity Utopia

employment grew. New industries emerged that nobody could have predicted.

The WEF’s projection of 92 million jobs displaced and 170 million jobs created by 2030 is consistent with this historical pattern. The Yale-Brookings finding of no aggregate employment shock three years after ChatGPT’s launch is consistent with it. The data showing 88% of workers already using AI without mass layoffs is consistent with it.

This does not mean no one will lose their job.

People will lose jobs. Some already have. But the idea that AI will create a world without work is not supported by evidence — and believing it leads to the worst possible response: passive defeatism. If you think the robots are coming no matter what, you don’t learn the tools. You don’t adapt. You wait for UBI or a political solution. And while you wait, the person next to you is getting better at using AI, capturing the gains, and outcompeting you.

NARRATIVE TWO: THE PRODUCTIVITY UTOPIA The second narrative says AI will make everyone so productive that work becomes optional, quality of life will skyrocket, and we’ll all enjoy four-day workweeks doing creative, fulfilling things. This story is also wrong — not because the productivity gains aren’t real, but because it ignores who captures them.

REPLACED 23

The real threat is neither apocalypse nor utopia

Productivity gains do not automatically accrue to workers. They accrue to whoever has the bargaining power to capture them. In a competitive labor market where augmented workers are plentiful, the gains flow to employers and consumers in the form of lower prices and higher profits. In a labor market where augmented workers are scarce — which is the current moment, given only 5% deep adoption — the gains flow disproportionately to the augmented workers themselves, in the form of higher wages, better positions, and more opportunity. The window where augmented workers are scarce will not stay open forever.

The productivity utopia narrative is dangerous because it says “don’t worry, everyone benefits.” Not everyone benefits — not automatically, not without deliberate effort on your part.

THE REAL THREAT IS NEITHER APOCALYPSE NOR UTOPIA The real threat is something much more mundane, and for that reason much more important:

redistribution.

AI is redistributing competitive advantage between workers, right now, inside your company, inside your industry, inside your profession. It is redistributing it from people who work the old way to people who have learned to multiply themselves with the new tools. It is REPLACED 24

redistributing it slowly enough that you can miss it week to week, but fast enough that if you look up in eighteen months, the landscape of your profession will be different.

You can already see this if you know where to look:

The manager who used to need three direct reports and now manages effectively with one augmented report.

The freelancer who used to compete against ten local rivals and now competes against a hundred global rivals, including some who use AI to deliver in days what used to take weeks.

The recent graduate with AI fluency who outperforms mid-level employees with three times their tenure.

The founder who runs a seven-figure business with a team of three and an AI stack.

The writer who publishes three times the volume and wins clients away from slower competitors.

The analyst who can process ten times the data and produce insights that win promotions.

The designer who iterates faster, explores more options, and wins more pitches.

The coder who ships features at a pace that makes their peers look slow by comparison.

REPLACED 25

The greatest danger: being average

These are not apocalyptic scenarios. They are quiet, continuous, cumulative shifts. They happen one project at a time, one promotion at a time, one hiring decision at a time, one client at a time. And after a few quarters, the people who have adapted are earning more, producing more, and occupying better positions — and the people who haven’t are wondering why they’re getting passed over.

THE GREATEST DANGER: BEING AVERAGE The single most dangerous position to occupy in the next five years is not “working in a job that AI can do.”

Very few entire jobs will disappear. The most dangerous position is being an average performer doing average work in a world where an augmented performer can produce above-average output faster and cheaper than you.

When employers have access to augmented workers who deliver 1.3-2x the output at the same cost, average becomes expensive. Average becomes unnecessary. Average becomes the target for layoffs, for stagnation, for being stuck.

This is not a moral judgment. It is not about whether you are a good person or a hard worker. It is a market reality. And it is a reality you can respond to.

The good news — and it is very good news — is that this is still early. The 5% number means you have not REPLACED 26

missed the window. The fact that most people are still using AI casually means the competitive moat for serious adopters is still wide. The fact that organizations have not yet redesigned their workflows around AI means the biggest gains are ahead of us, and the people who learn to work with AI now will be positioned to lead those redesigns.

But the window is finite. Every month that passes, more people wake up. More companies implement AI training. More workers begin the shift from casual use to deep integration. The early-adopter advantage gets a little smaller. The floor gets a little higher.

The question is not whether AI will change work. It already has. The question is whether, when the change completes itself in your corner of the professional world, you are the one holding the advantage — or the one it is being held over.

REPLACED 27

PART II: THE DIVIDE

PART II: THE DIVIDE Chapter 5: The New Productivity Gap For most of the twentieth century, the gap between a mediocre worker and an excellent worker in most fields was bounded. A top salesperson might outsell an average one by 50%. A star software engineer might be 2x as productive as an average engineer. A top marketing copywriter might produce better work, but not 3x as much. Talent mattered, effort mattered, but the ceiling on individual productivity was constrained by human limits on time, attention, and cognitive throughput.

AI shatters that ceiling.

Not because AI makes people smarter in some fundamental sense. But because it removes the bottleneck of throughput. A worker who knows how to delegate cognitive work to AI well can iterate more times, explore more options, process more information, draft more alternatives, and cover more ground in a day than a worker who does everything by hand. The gap between an AI-fluent worker and an AI-

resistant worker doing the same job is no longer 20%

or 50%. In some domains it is 2x, 3x, even 5x.

REPLACED 28

The new spread in each field

Let me make this concrete.

THE NEW SPREAD IN EACH FIELD Writing and content. A writer without AI might produce 1,500 words of polished content per day. A writer using AI well — not accepting raw output, but using it to generate drafts, variations, angles, research summaries, and edits — can produce 4,000-6,000 words of comparable or better quality per day. That is not a 10% difference. That is a structural difference in output capacity.

Coding. Peng et al. found a 26% completion rate increase, but anecdotal reports from senior engineers fluent with modern coding assistants suggest 2-3x throughput on routine work and significant acceleration on hard problems. The best engineers using AI well are the equivalent of small teams.

Analysis and consulting. The BCG study showed 25% faster completion with better quality on in-range tasks. For skilled analysts who know how to structure problems, validate outputs, and use AI as a research and calculation engine, the multiplier on analytical throughput is even larger — not because AI provides answers, but because it removes the friction of data processing, chart generation, literature review, and draft preparation.

Design. A designer using AI image generation for iteration can explore 20 visual directions in the time a REPLACED 29

The gap is widening, not shrinking

traditional designer explores 3. That doesn’t replace taste — the designer still selects, directs, refines —

but it multiplies the breadth of exploration. Clients can see more options faster. Designers who can iterate faster win more pitches.

Customer service and sales. The Brynjolfsson study showed 14% average, 34% novice gains.

Salespeople using AI to research prospects, draft personalized outreach, prepare call summaries, and follow up can manage more relationships simultaneously than salespeople working manually.

Legal work. Junior associates used to spend hundreds of hours reviewing documents in discovery.

AI-assisted review is faster, cheaper, and increasingly standard. Associates who can supervise AI review and focus on higher-level legal judgment outperform those who can only do manual review.

THE GAP IS WIDENING, NOT SHRINKING There are two reasons this gap will widen before it narrows.

First, AI tools are improving. The models are getting better at more tasks. New tools appear every week. Someone who invested in learning AI tools six months ago has an advantage today, but the advantage compounds as the tools improve and they learn to use the new capabilities. A person who hasn’t REPLACED 30

What this does to compensation

started is not just six months behind; they’re falling further behind every month.

Second, skill with AI compounds with domain expertise. A mediocre writer using AI can produce mediocre content faster. A strong writer using AI can produce excellent content faster. The gap between mediocre-fast and excellent-fast is wider, not narrower, than the old gap between mediocre-slow and excellent-slow. The amplifying effect of AI rewards depth — because you have to know what good looks like to curate AI output effectively.

WHAT THIS DOES TO COMPENSATION The PwC finding of a 56% wage premium for AI-skilled workers sounds dramatic. It is not surprising. In a market where some workers produce 2x what their peers produce, employers will compete for those workers. Freelance marketplaces already show AI-

fluent specialists charging premium rates because they deliver faster without sacrificing quality.

But here is what most people miss: the premium is not for knowing how to prompt an AI. Prompting is trivial. The premium is for the combination of domain expertise, good judgment, and AI fluency that lets someone produce augmented output at quality levels the un-augmented cannot match. AI fluency alone is not enough. Domain expertise alone is no longer REPLACED 31

The window before the floor rises

enough. It’s the combination that commands the premium.

THE WINDOW BEFORE THE FLOOR RISES Right now, we are in a window where: - AI tools are powerful enough to deliver real productivity gains -

Most people are not using them well - Most organizations have not redesigned roles, compensation, or workflows around AI yet - The skills gap is large enough that AI-fluent workers are scarce This is the window where you can build an outsized advantage. It will not stay open forever. As more people develop AI fluency, the floor rises. What is today a 56% premium becomes tomorrow’s baseline expectation. The people who are already fluent by then will be the ones setting the standards, leading teams, capturing senior roles. The people who wait until AI fluency is mandatory will find that they are competing in a more crowded market, doing the same thing as everyone else, with no advantage.

The time to move is now.

REPLACED 32

Chapter 6: Who Wins, Who Loses, and Why

Chapter 6: Who Wins, Who Loses, and Why Who wins in an AI-augmented labor market? Who loses? After three years of data, the patterns are clear enough to describe. Let me draw them as precisely as I can.

THE WINNERS 1. The AI-fluent expert. This is the single most advantaged position in the current market. Someone who already has deep domain expertise — the lawyer who understands complex transactions, the marketer with twenty years of brand-building intuition, the engineer who knows systems architecture, the doctor with diagnostic judgment, the salesperson who knows how to read a room — and who adds AI fluency to that expertise. AI multiplies their existing judgment, allowing them to produce more output, serve more clients, tackle more complex problems, and move faster than their peers. They capture the PwC 56%

wage premium. They get promoted. They start businesses. They consult at premium rates. They are the clearest winners.

2. The fast-learning novice. Because AI compresses the learning curve — because it can transfer tacit knowledge from experts in real time — a smart, motivated newcomer can become productive REPLACED 33

faster than at any point in history. The 34% gain for novice support agents is the leading indicator. A new graduate who enters a field already fluent with AI tools, who learns domain fundamentals while leaning on AI for execution, can close a multi-year experience gap in months. These workers threaten mid-level incumbents who rely on routine competence.

3. The entrepreneur / solo operator. This is where the leverage is most dramatic. A single founder with AI tools can now do work that used to require teams: marketing copy, design mockups, data analysis, customer support scripts, code, video editing, financial models, legal documents. The minimum viable team has gotten smaller. Solo founders are building businesses at revenue levels that previously required 5-10 employees. The founder who combines business judgment with AI leverage captures almost all the surplus.

4. The orchestrator. The manager who knows how to combine AI tools and human teams — who can design workflows where AI handles the routine and humans handle judgment, who can validate AI output, who can translate business problems into problems AI can solve — becomes dramatically more valuable than a traditional manager. Organizations will pay for people who can deliver AI-augmented results at the team level, not just the individual level.

REPLACED 34

The losers

5. The builder. The people building the AI tools themselves — AI engineers, applied scientists, ML infrastructure engineers — remain in extreme demand. But “builder” here also includes the people who build AI-powered workflows and systems inside non-tech organizations, the people who integrate AI into business processes, the people who can set up and maintain AI systems for teams that don’t have that expertise in-house. This is a layer of the economy that is still being built.

6. The human connector. AI cannot build trust, read a room, navigate political dynamics, hold a difficult conversation with empathy, negotiate a delicate deal in person, or be present with a client in crisis. The people who are genuinely excellent at high-

stakes human relationship work — top-tier sales, leadership, counseling, negotiation, coaching, key account management, politics — will see their value hold and even rise, because AI can’t do it.

THE LOSERS 1. The routine knowledge worker in denial. If you spend most of your day producing structured, predictable cognitive output — standard reports, routine emails, first-draft analysis, basic code, basic design, standard slide decks, data entry, basic content creation — and you have made no effort to learn AI tools or move up the value chain, you are the most exposed. You are competing against both augmented REPLACED 35

novices and augmented experts, and your position is eroding from both sides.

2. The “prompt engineer” with no domain expertise. Knowing how to write prompts is not a durable advantage. As the models get better at understanding instructions, the skill of prompting becomes trivial — like knowing how to use a search engine. The person who can only write prompts but has no domain expertise, no judgment, no taste, no relationships, will find their skills commoditized fast.

AI fluency without underlying substance is a temporary advantage at best.

3. The over-truster. The person who believes AI output without verifying, who lets AI make decisions that require human judgment, who outsources their critical thinking — this person will eventually make costly mistakes. The BCG study showed that AI users on hard problems were 19 percentage points worse than non-users because they trusted confident wrong answers. In the long run, over-trusters get fired, sued, or humiliated. AI is a tool, not an authority. People who forget this lose.

4. The pure resister. The person who refuses to use AI on principle, who insists the old ways are better, who dismisses AI as a fad or a fraud — this person is the equivalent of the secretary who refused to learn word processing in 1985. They will not be fired tomorrow. They will be fired three years from REPLACED 36

The middle path

now when they are the only person in the department taking three times as long to produce inferior output.

5. The person who confuses activity with value.

AI creates the possibility of enormous volumes of output. Masses of emails, drafts, code, content, reports, and analysis can now be produced quickly.

But producing more volume is not the same as producing more value. The person who uses AI to generate huge amounts of low-value output — the person who fills inboxes with AI-written emails, floods platforms with AI-written content, buries colleagues in AI-generated analysis — is adding noise, not value.

Organizations will not reward noise. They will reward people who use AI to produce better outcomes, not more volume.

6. The burned-out chaser. A small but real group of people will chase AI tools so frantically that they never develop depth. They try every new model, every new platform, every new framework, and never sit still long enough to build real judgment in a domain. They become knowledgeable about tools but ineffective at work. Tool-chasing is a form of procrastination.

Mastery of a few tools integrated into a real workflow beats superficial knowledge of fifty tools.

THE MIDDLE PATH Where do most people actually fall? Not in the “winners” column yet, not in the “losers” column yet.

REPLACED 37

Chapter 7: The Skills AI Can’t Touch — And the Skills It Makes More Valuable

Most people are in the middle. They have tried AI.

They use it for email and summarization. They have not redesigned their work around it. They are not yet in danger, but they are not yet positioned to win.

The entire point of this book is to move you from the middle to the winners column — deliberately, systematically, without gimmicks or get-ahead-quick schemes. The rest of the book shows you how.

Chapter 7: The Skills AI Can’t Touch — And the Skills It Makes More Valuable If you want to know where to invest your time, you need to know which skills are safe and which skills AI makes more valuable. Let me give you a framework that I use myself and that I recommend to everyone I mentor.

There are three categories:

Category 1: Skills AI largely replaces. These include routine cognitive tasks: basic summarization, first-draft writing, standard formatting, basic translation, straightforward data entry, simple coding, template-based design, basic research, predictable analysis, standard slide creation. If your workday is primarily made of these tasks, your position is exposed. You should not spend time trying to get REPLACED 38

better at these tasks as a human — you should learn to use AI to do them faster, and redirect your time toward Category 2 and Category 3 skills.

Category 2: Skills AI makes more valuable.

These are skills where having AI as an assistant dramatically increases the leverage of the human who has them. They include: problem-framing (knowing which question to ask), taste and aesthetic judgment (knowing which output is good), strategic decision-

making (choosing between alternatives), project scoping (defining what needs to be built), verification (catching AI’s mistakes), editing (improving AI output to a professional standard), integration across domains (combining ideas from different fields), system design (composing workflows that combine multiple tools and people), teaching and mentorship (transmitting judgment, not just information), and client relationships built on trust. If you develop these skills, AI multiplies your value.

Category 3: Skills AI cannot meaningfully perform. These are skills rooted in human particularity: building deep trust over time, demonstrating moral courage in difficult decisions, exercising empathy in situations of real vulnerability, leading people through change and loss, creative originality at the highest level (the kind that redefines a field rather than remixes existing patterns), negotiating in situations with genuine conflict and emotion, physical presence and charisma, ethical REPLACED 39

judgment where real people’s lives are at stake, cultural fluency that comes from lived experience in a community, and accountability (someone who takes responsibility when things go wrong). These skills are not touched by AI — and in many ways, as more work becomes AI-mediated, the premium on genuine human presence and character goes up.

Let me go deeper on a few of the most important.

Judgment. AI can produce options. It can present analysis. It can list pros and cons. It cannot decide which choice is right — not in the sense of optimal, not when the variables include organizational politics, human emotion, long-term reputation, ethical trade-

offs, your own values, your sense of what is worth doing. Judgment is the result of experience, reflection, failure, character, and a sense of context. It is the rarest and most valuable skill in business and life, and AI makes it more important because it increases the volume of decisions you have to make and the speed at which you have to make them.

Taste. If you produce creative or design work —

writing, marketing, product, visual design, strategy —

AI can generate dozens of options. But deciding which option is genuinely good, which has the ring of truth, which will resonate with an audience, which is original versus derivative, which aligns with the taste and identity of your brand or client — that is taste. People with taste use AI to explore more options faster.

REPLACED 40

People without taste get overwhelmed by AI output or select mediocre results. Taste is developed by years of looking at, reading, and consuming the best work in your field. It cannot be faked. It cannot be automated.

It is the filter that turns AI-generated volume into quality.

Verification. Every experienced user of AI has a story about an AI hallucination — a fake case citation, a made-up statistic, a plausible-sounding but completely wrong analysis, a code suggestion that looks clean but introduces a bug. The people who get real value from AI are people who verify everything, who cross-check claims, who know when AI is likely to be wrong, who have enough domain knowledge to spot errors. People who can’t verify AI output are dangerous. They will confidently produce and propagate wrong information. People who can verify become the essential bottleneck through which AI output must pass.

Problem-framing. AI solves problems extremely well — when the problem is correctly framed. But most of the value in any complex project comes before the problem-solving starts. Which problem are we actually trying to solve? What are we measuring?

What does success look like? Who are the stakeholders? What constraints actually matter? The person who frames the problem well will get excellent work from AI. The person who frames it poorly gets polished answers to the wrong question.

REPLACED 41

Chapter 8: The Five Traps That Get Good People Fired First

Relationship-building and trust. AI cannot build trust over time. It cannot make someone feel genuinely seen and heard in a difficult conversation. It cannot read a room. It cannot look a client in the eye and earn their confidence over years. In a world where first-draft communication can be AI-generated, the human gesture — the handwritten note, the personal phone call, the genuine moment of connection — becomes more, not less, valuable.

Ownership and accountability. When something goes wrong, AI cannot take responsibility. It cannot be fired. It cannot feel remorse. It cannot make amends.

Organizations run on accountability. The person who takes ownership — of a project, of a failure, of a team’s outcome — is irreplaceable. AI is a tool in the hands of an owner.

The play for the next decade is clear: automate Category 1 skills using AI, invest aggressively in Category 2 skills, and deepen Category 3 skills throughout your life and career. People who do all three are the people who will thrive. CH7 echo “Added chapter 7”

Chapter 8: The Five Traps That Get Good People Fired First After talking to hundreds of workers, managers, and founders navigating the AI transition, I see the same REPLACED 42

Trap 1: The “It Won’t Happen to Me” Trap

mistakes repeated. These traps are dangerous precisely because they are tempting. They feel reasonable in the moment. But they are the patterns that separate people who thrive from people who get pushed aside.

TRAP 1: THE “IT WON’T HAPPEN TO ME” TRAP The most dangerous phrase in any technology transition is “My job is different.” People always believe their work is too nuanced, too creative, too dependent on human judgment, too relationship-based to be affected. The radiologists said it about computer-

aided diagnosis. The travel agents said it about online booking. The retail clerks said it about e-commerce.

The typesetters said it about desktop publishing. They were all partially right — their work changed, it was not eliminated — but they were also partially wrong:

the people who adapted kept their jobs, and the people who didn’t were displaced.

Assume AI will affect your work. Assume it will change how your job is done, how many people are needed to do it, and what skills are rewarded. The only question is whether you adapt proactively or reactively.

REPLACED 43

Trap 2: The Tool Collector Trap

TRAP 2: THE TOOL COLLECTOR TRAP There is a personality type that responds to AI by collecting tools. They have accounts with ten different AI platforms. They know the latest model release.

They read every newsletter. They argue on Twitter about which LLM is best. They can demonstrate a hundred impressive tricks. But they never integrate any of this into their actual daily work. Their output doesn’t improve. Their productivity doesn’t rise. They can talk about AI for hours, but they don’t use it to produce anything of greater value.

Do not confuse knowing about tools with being able to do work with tools. Master one or two core tools deeply and integrate them into your workflow before you chase the next release.

TRAP 3: THE OVER-TRUST TRAP The BCG study showed it clearly: on hard problems, AI users produced 19% worse results because they trusted confident wrong answers. The more fluent you become with AI, the easier it is to slide into over-trust — to skim its output instead of reading carefully, to accept its analysis without verification, to let it make judgment calls that should be yours.

This trap gets people fired. It gets consultants sued.

It gets engineers deploying buggy code. It gets analysts presenting wrong numbers. It gets managers making bad decisions.

REPLACED 44

Trap 4: The “I’m Too Senior for This” Trap

Make verification a habit. Cross-check numbers.

Read important AI output line by line. Never let AI make a consequential decision on its own.

TRAP 4: THE “I’M TOO SENIOR FOR THIS” TRAP Some experienced workers — especially those with a decade or more in their field — respond to AI with a version of “I built my career without this, I don’t need to learn it now.” They expect juniors to handle the tools. They think AI is for entry-level work.

This is catastrophic. In an AI-augmented organization, the people who direct the tools are the people who understand the tools. If you can’t personally use AI effectively, you can’t manage people who do — you can’t evaluate their output, you can’t design the workflow, you can’t spot mistakes, you can’t tell what’s hard versus easy. Senior people who don’t learn the tools become dependent on junior people who do. That dependency erodes authority and relevance. Seniority plus AI fluency is the most powerful combination in the market. Seniority without AI fluency is a liability.

TRAP 5: THE “AI WILL DO THE WORK FOR ME” TRAP Some people respond to AI by stopping work. They let AI write their emails without editing. They let AI REPLACED 45

Chapter 9: The Hidden Cost of Outsourcing Your Thinking

produce their reports without thinking. They let AI design their presentations without crafting the story.

They stop generating their own ideas and start accepting AI’s first suggestion. Their work becomes generic, flat, AI-flavored. Colleagues and clients can tell. It has a polished sameness to it — competent on the surface, but without edge, without voice, without originality.

AI is an amplifier, not a replacement for thinking.

The people who get the best results work harder than ever when using AI — thinking harder about what to ask, iterating more times, editing more carefully, directing the output toward a specific vision they hold in their head. They don’t think less. They think better, because the cognitive load of routine work is lifted and they can focus on higher-order decisions.

Avoid these five traps. They sound obvious when you read them. They are much harder to avoid in practice, especially when AI makes things easy and fast and you’re busy.

Chapter 9: The Hidden Cost of Outsourcing Your Thinking There is a cost to using AI that almost nobody warns you about. It is not bias, it is not hallucination, it is not job displacement — though all three matter. It is something quieter and more insidious.

REPLACED 46

Cognitive atrophy.

If you stop performing certain kinds of thinking because AI does them for you, your ability to perform that thinking atrophies. It is the same principle as muscle mass: use it or lose it.

A 2025 study from MIT’s Media Lab found early evidence of this effect. In writing tasks where participants repeatedly relied on AI for ideation and drafting, they performed worse on subsequent unaided writing and creative-thinking tasks compared to participants who had practiced without AI.

Researchers dubbed it the “cognitive cost” of AI reliance. The effect was small in the short term but compounded over time.

This is consistent with a much older body of research on automation — from aviation to medicine to the military — where automation bias, skill fade, and complacency are well-documented. When a system routinely does a task for you, your ability to monitor that system degrades. When the system fails — and all systems fail — you are less prepared to step back in.

Pilots who rely too heavily on autopilot lose their manual flying reflexes. Radiologists who rely too heavily on AI detection lose sensitivity to edge cases.

Writers who let AI do all their drafting lose their voice.

Analysts who let AI do all their number-crunching lose REPLACED 47

The principle of deliberate non-delegation

their intuition for what the numbers should look like, which means they catch fewer errors.

Let me be careful not to overstate this. The cognitive atrophy risk is real, but it is not an argument against using AI. It is an argument for using AI thoughtfully — in ways that preserve and strengthen your core skills rather than letting those skills decay.

THE PRINCIPLE OF DELIBERATE NON-

DELEGATION The solution is not to reject AI. The solution is to be deliberate about what you delegate and what you don’t.

Here is a simple rule I recommend to everyone I coach: for each category of cognitive task you perform, decide explicitly:

DELEGATE FULLY: Tasks you have already mastered, where your judgment is strong enough to verify the output, and where doing it yourself would waste time you should spend on higher-value work. Example: formatting a document, scheduling a routine meeting, summarizing a long report you already understand.

REPLACED 48

The mental workout principle

DELEGATE DRAFTS, NOT DECISIONS:

Tasks that involve judgment, voice, or creativity — have AI produce a draft, then edit, rewrite, and improve it yourself. You get speed without surrendering authorship. Example: writing an important email, drafting a presentation, analyzing data, writing code. This is where most knowledge work should live.

DO NOT DELEGATE: Tasks that build or maintain skills you cannot afford to lose; tasks that involve moral judgment or high-stakes decisions; tasks where your voice, taste, or presence is the value. Example: writing anything under your name that people judge you by; giving critical feedback to a direct report; making decisions that affect people’s careers; creative work that represents your core artistic vision; final verification of anything that will be sent to a client or boss.

THE MENTAL WORKOUT PRINCIPLE Think of your cognitive skills the way a serious athlete thinks about physical training. An athlete uses specialized equipment — better shoes, a lighter bike, a power meter — to improve performance. But they don’t stop running, cycling, or lifting just because the equipment is better. They train with the equipment to get stronger.

REPLACED 49

Use AI the same way. Use it to amplify your performance, but deliberately practice your core cognitive skills without AI on a regular basis: write something important from scratch without AI drafts, work through a hard problem with pen and paper, do a calculation manually to develop intuition, read a long book without AI summarization. Keep your cognitive machinery in shape.

The people who will be most valuable in an AI-rich world are not the ones who let AI do all their thinking.

They are the ones who use AI to extend their reach while deliberately maintaining the core human skills of reasoning, judgment, creativity, and presence.

REPLACED 50

PART III: THE PLAYBOOK

PART III: THE PLAYBOOK Chapter 10: The AUGMENT Framework — Six Moves That Separate Thrivers from Survivors After studying how the most effective AI-augmented workers actually operate — across software, law, consulting, writing, design, sales, education, and entrepreneurship — I have identified six practices that separate people who are thriving with AI from people who are being left behind.

They form the acronym AUGMENT:

A — Audit your work U — Up-skill selectively G — Generate systems, not just outputs M — Maintain your human edge E — Elevate your visibility N — Negotiate your leverage T — Tighten your safety net •

REPLACED 51

A — Audit your work

These are not gimmicks. They are not prompt engineering tricks. They are six durable practices that will serve you regardless of which AI models dominate a year from now or five years from now.

Let me walk through each one.

A — AUDIT YOUR WORK Before you can use AI effectively, you need to know what your work actually consists of. Most people have never systematically mapped what they do each week.

They have a vague sense — “emails, meetings, reports, projects” — but they don’t know how their time breaks down, which tasks create the most value, and which tasks are routine enough to delegate to AI.

Here is the exercise:

For the next five working days, keep a simple log.

At the end of every two-hour block, write down: 1.

What you did 2. How long it took 3. Whether it was routine / semi-routine / high-judgment / creative /

relational 4. Whether AI could have done any part of it 5. Whether it created real value for your organization, client, or career At the end of the week, look at the results. You will likely discover that: - A surprising share of your time is spent on Category 1 routine tasks that AI can accelerate dramatically - Some of your highest-value work (relationships, judgment, strategy) takes less REPLACED 52

U — Up-skill selectively

time than you think, crowded out by routine - You have “time drains” that neither AI nor anything else justifies Once you have this map, you know where to apply AI first. Start with the biggest block of routine work that takes the most time and creates the least differentiation value. Freeing those hours gives you room to invest in Category 2 and 3 work.

Do this audit quarterly. AI tools change. Your work changes. A map from six months ago may already be out of date.

U — UP-SKILL SELECTIVELY The biggest mistake people make with AI learning is trying to learn everything. Don’t. Be ruthlessly selective.

You need three layers of skill:

Foundation layer: AI literacy. Understand what current AI can and cannot do. Learn the strengths and failure modes of one or two leading models. This takes maybe five hours total — not five months. You don’t need a computer science degree. You need to use the tools, read a few honest analyses of their limits, and develop a gut sense for when AI is reliable and when it isn’t.

REPLACED 53

Workflow layer: 2-3 tools that matter for your job. Don’t collect twenty tools. Identify the two or three AI tools that most directly impact your core work — probably a general-purpose LLM for drafting and analysis, possibly one domain-specific tool (a coding assistant if you’re a developer, an image generator if you’re in creative work, a data analysis tool if you’re in analytics), and learn them deeply.

Learn their shortcuts, their edge cases, how to get the best output, how to chain them with your existing software. Mastery of two or three tools beats casual acquaintance with twenty.

Judgment layer: the skills AI can’t replace. This is where 70% of your development time should go.

Deepen your domain expertise. Improve your writing.

Hone your taste. Develop your leadership. Learn to facilitate better meetings. Get better at reading people. Read books in your field. Study the history of your industry. Build your network. These investments never become obsolete — AI makes them more valuable.

The ratio is important. If you spend 10% of your development time on AI tools and 70% on human skills and domain depth, with 20% on integration, you will outperform someone who spends 80% of their time chasing new AI tools and 20% on substance.

REPLACED 54

G — Generate systems, not just outputs

G — GENERATE SYSTEMS, NOT JUST OUTPUTS Casual AI users ask AI to produce one output at a time: “write this email,” “summarize this document,”

“draft this report.” That’s useful, but it’s a marginal improvement.

Serious AI users build systems. They design repeatable workflows where AI handles categories of work consistently. Examples:

Instead of asking AI to draft one customer follow-up email, create a set of reusable templates that AI fills in from your notes after every call.

Instead of using AI to answer one research question, build a personal knowledge base where AI can answer recurring questions from your own documents and notes.

Instead of manually preparing a weekly status report each Friday, set up a workflow where AI compiles the week’s activity from your calendar, messages, and task list into a draft you edit and send.

Instead of using AI to debug one piece of code, configure your coding assistant to catch recurring categories of bugs automatically.

REPLACED 55

M — Maintain your human edge

Instead of manually formatting presentation slides each time, create a template workflow where AI drafts slides from a structured outline you provide.

Systems compound. A system that saves you 30

minutes a week becomes 26 hours a year. A system that saves your team 30 minutes a week multiplies by the number of people. Over a year, people who build systems pull far ahead of people who use AI for one-off tasks.

Start small. Identify one recurring task you perform at least weekly. Build a repeatable AI-assisted workflow for it. Refine it until it works reliably. Then move to the next one. After a year, you will have a personal operating system that makes you dramatically more productive than your peers.

M — MAINTAIN YOUR HUMAN EDGE As AI handles more cognitive routine, the parts of work that are human become more valuable, not less.

Double down on them deliberately.

Be the person in the room. As remote work and AI-mediated communication grow easier, physical presence becomes a differentiator. Show up. Look people in the eye. Be at the meeting, the dinner, the event. The relationships built in person are harder to replicate.

REPLACED 56

E — Elevate your visibility

Develop a point of view. AI can summarize conventional wisdom. It cannot have an original, lived perspective rooted in your experience, your values, and your observation. Form opinions. Take positions (held lightly, with room to update). Write or speak about what you believe. People follow people who think for themselves.

Build trust deliberately. Trust is built over time through consistent follow-through, honesty, discretion, and care. AI cannot build trust for you. Every kept promise, every difficult conversation handled well, every piece of honest feedback delivered with empathy, builds a reservoir that compounds over a career.

Stay physically and mentally healthy. This sounds trivial. It is not. The cognitive performance gap between a well-rested, focused, emotionally grounded person and a tired, anxious, distracted person is enormous — and AI won’t close it. Sleep, exercise, nutrition, relationships, and quiet time are not luxuries. They are the foundation of high performance in an age where cognitive work is the competitive arena.

E — ELEVATE YOUR VISIBILITY Being good at your work with AI is not enough if nobody knows. In a rapidly changing environment, the REPLACED 57

people who get promoted, hired, and recruited are the people whose augmented work is visible.

Document and share what you’re learning.

Write short internal posts about how you’re using AI to improve work in your team. Share useful workflows with colleagues. Give a lunch-and-learn on AI in your role. People who teach AI practices become recognized as leaders in the new way of working.

Capture your augmented wins. When AI helps you deliver something faster, better, or cheaper, document it. Not for bragging — for your next performance review, your resume, your LinkedIn.

Quantify the gain: “Reduced report preparation time from 8 hours to 2 hours while improving quality, using AI-assisted research and drafting.” Specific, measured wins are what promotions and job offers are built on.

Speak publicly about your domain, not just about AI. Write, post, speak, create content about your area of expertise — showing, not telling, that AI is making your thinking sharper, not replacing it.

Don’t become an “AI person” unless AI is your field.

Become a person in your field who is visibly mastering the new tools.

Make your work product recognizable. Generic AI output looks like everyone else’s generic AI output.

Develop a distinctive voice, style, and approach that makes your work identifiable as yours — even when REPLACED 58

N — Negotiate your leverage

you use AI as an assistant. That distinctiveness is your brand.

N — NEGOTIATE YOUR LEVERAGE As you become more productive using AI, your leverage with your employer and in the market increases. Use it deliberately.

Track your augmented productivity. If AI has let you take on more work, serve more clients, or deliver faster, document that. In your next performance review, come with data. Negotiate from evidence, not from hope.

Don’t give away all your productivity gains for free. If AI makes you 30% more productive, some of that surplus should accrue to you — in compensation, promotion, reduced hours, flexibility, equity, or the freedom to choose higher-value projects. People who quietly produce more without negotiating create a situation where their employer captures 100% of the surplus and expects the new level of output as baseline.

Develop external options. The strongest negotiating position is being able to leave. Maintain your network. Keep your resume current. Talk to recruiters occasionally. Build a side income stream if you can. The person who has no option but to stay is the person who gets the worst deal.

REPLACED 59

T — Tighten your safety net

Capture equity, not just wages, where possible.

If you’re in a position to take equity — in a startup, in your employer through stock options, in your own business — remember that productivity gains from AI flow disproportionately to owners of capital, not just labor. Being an owner, even in a small way, changes your long-term trajectory.

T — TIGHTEN YOUR SAFETY NET Even if you do everything right, transitions happen.

Companies lay off good people. Industries restructure.

AI changes faster than some people can adapt. Don’t be caught unprepared.

Build a financial runway. Aim for six to twelve months of living expenses in savings. This is not pessimism — it is freedom. A runway lets you weather layoffs, take time to retrain, negotiate from strength, or start something new.

Maintain an active professional network. Don’t wait until you need a job to talk to people. Check in with former colleagues. Attend industry events. Help people when you can. A strong network is the single best insurance against career disruption.

Keep learning continuously. Skills depreciate faster in the AI era. Dedicate at least five hours a week to deliberate learning — some on AI tools, most on domain skills and human skills. Build learning into REPLACED 60

Chapter 11: Your First 30 Days: From AI-Curious to AI-Competent

your routine, not something you do in a panic when a job is at risk.

Cultivate multiple income streams if you can.

This doesn’t mean quitting your job to freelance tomorrow. It means developing options: a side consulting practice, a product you sell online, content that generates passive income, teaching or coaching.

Diversification reduces risk.

AUGMENT is not a one-time project. It is a set of habits. Audit quarterly. Learn continuously. Build systems incrementally. Maintain your edge. Elevate your visibility. Negotiate from strength. Tighten your net. The people who treat this as ongoing practice will be the people who thrive. CH10 echo “Added chapter 10 AUGMENT”

Chapter 11: Your First 30 Days:

From AI-Curious to AI-

Competent You do not need to become an AI expert in a month.

You do not need to learn to code. You do not need to buy a dozen tools. What you need in your first 30 days is to go from “I’ve tried AI a few times” to “AI is a reliable part of how I work, every single day.”

Here is a practical 30-day plan.

REPLACED 61

Week 1: Set up your toolkit and run your audit

WEEK 1: SET UP YOUR TOOLKIT AND RUN YOUR AUDIT Day 1: Pick your primary tool. Choose one leading general-purpose AI assistant and commit to it. Don’t agonize over which one — by the time you read this, the leading models will have changed anyway. Pick one with broad capabilities (writing, analysis, coding, image, research) and stick with it. Pay for a subscription. The free tiers are too limited for serious work.

Day 2: Pick one domain-specific tool. If you write code, install an AI coding assistant. If you work with data, learn an AI data tool. If you create visual content, get an AI design tool. If you’re in marketing or content, learn one content workflow tool. That’s it.

Two tools total for Week 1.

Days 3-5: Run your work audit. Use the framework in Chapter 10. Log your work in two-hour blocks for three working days. Identify three specific, recurring tasks that take significant time and are AI-

suitable. These become your first targets.

Weekend (Days 6-7): Pick one of the three tasks and write a short description of what you currently do manually. Then spend one hour experimenting with having AI do a first pass. Don’t worry about getting perfect output yet. Just get a feel for how to describe the task and what kinds of results you get back.

REPLACED 62

Week 2: Integrate one core task

WEEK 2: INTEGRATE ONE CORE TASK Day 8: Choose one of your three target tasks.

Make it the task you will AI-assist every day for the rest of the week. The key here is consistency.

Every time you do that task this week, you use AI for the first pass.

Day 9-12: Build your first prompt template.

Don’t start from scratch every time. Write a reusable prompt for your target task. Include: context (what you’re working on, who the audience is), the task (what you want AI to do), format (how you want the output structured), and constraints (tone, length, style). Save this template somewhere accessible — a notes app, a text file, a snippet manager. Refine it after every use based on what worked and what didn’t.

Day 13: Add your second task. Same process.

Integrate a second recurring task. Write a template.

Use it daily.

Day 14 (End of week 2): Review. After two weeks of using AI for two recurring tasks, assess: how much time did you save? What did you do with that time?

Where did AI produce good results? Where did it fail?

What did you learn about what to trust and what to verify?

REPLACED 63

Week 3: Build a system and start verifying

WEEK 3: BUILD A SYSTEM AND START VERIFYING Day 15: Turn one of your tasks into a system.

Instead of prompting from scratch every time, build a repeatable workflow. For example: “After every client call, paste my notes into this template; AI produces a summary, action items, and a follow-up email draft; I review and edit; I send the email.” Write the system down. Use it every time.

Day 16-18: Practice verification deliberately.

For every AI output you use this week, spend two minutes checking: are the facts correct? Is the tone right? Does it say anything I wouldn’t stand behind?

Are there hallucinations? Catching errors is a skill —

build it deliberately.

Day 19: Add your third core task. By now you should be using AI for three recurring tasks every day, with templates and a basic system for at least one.

Day 20-21: Learn one advanced feature. Maybe it’s custom instructions, maybe it’s uploading documents for analysis, maybe it’s image generation, maybe it’s connecting AI to your notes. Pick one feature that unlocks new capability and spend two hours learning it.

REPLACED 64

Week 4: Expand and reflect

WEEK 4: EXPAND AND REFLECT Day 22: Have a conversation about AI at work.

Talk to a colleague about what you’ve been trying.

Share a tip. Ask what they’re doing. Learning accelerates in community.

Day 23-25: Apply AI to one task you previously thought was “yours.” A piece of writing you care about. A presentation you need to give. A piece of analysis that matters. Use AI as a collaborator, not a crutch — brainstorm with it, get it to challenge your thinking, use it to generate alternatives. See what happens when you bring AI into your core work rather than just routine tasks.

Day 26-27: Document what you’ve learned.

Write a one-page note to yourself: what works, what doesn’t, which prompts are reliable, which tasks AI is good for, where you still need to be careful. This document becomes the foundation of your personal AI playbook. Update it monthly.

Day 28: Review your 30 days. Compare how you work now with how you worked a month ago. Which habits have stuck? Where has AI genuinely improved your output? Where has it saved time? Where has it created risks or extra work? What’s the next skill or workflow to tackle?

Day 29: Set a 90-day goal. Based on what you’ve learned, set one concrete goal for the next 60 days: “I REPLACED 65

Chapter 12: Rewiring Your Daily Work: Before, With, and After AI

will reduce time spent on routine reporting by 50%,”

“I will publish one AI-assisted piece of content per week,” “I will use AI to handle first-draft coding so I can focus on architecture.” Make it specific and measurable.

Day 30: Teach someone. The best way to consolidate your learning is to teach it. Show a colleague one trick. Help a friend set up a workflow.

Teaching forces you to understand what you actually know.

After 30 days, you will not be an AI expert. But you will be past the beginner phase. You will have working habits. You will have your own data on what works for you. You will be lightyears ahead of the people who are still just “curious” about AI.

Then keep going.

Chapter 12: Rewiring Your Daily Work: Before, With, and After AI The most practical question people ask is: “What does an AI-augmented workday actually look like?”

Not the “I use AI for everything and never think anymore” extreme. Not the “I use AI once a week to write an email” extreme. The real thing — the pattern REPLACED 66

Before AI: You set the direction

used by people who are genuinely getting a 30-50%

productivity boost without sacrificing quality.

It looks like three phases: Before AI, With AI, After AI.

BEFORE AI: YOU SET THE DIRECTION The most important work with AI happens before you type a single prompt.

1. Know what you’re trying to produce. Before opening an AI tool, answer these questions in your head or in a quick note: - What am I trying to create?

(An email, an analysis, a draft, a plan, a decision?) -

Who is it for? (Boss, client, team, public, yourself?) -

What does good look like? (What are the criteria by which this output will be judged?) - What format should it take? (Length, structure, tone?) - What do I already know or believe about this?

People who skip this step get generic AI sludge.

People who do this step get useful AI output. The quality of AI’s work is almost entirely determined by the clarity of the human framing.

2. Gather context. What information does AI need that it doesn’t have? This might include: recent company decisions, specific data points, audience context, your own previous notes on the topic, style preferences, constraints. Provide context proactively.

REPLACED 67

With AI: You direct the work

AI is not omniscient. It doesn’t know what happened in your meeting yesterday unless you tell it.

3. Decide what you will not delegate. For this specific task, what parts must remain yours? If you’re writing an important email, you might decide the opening and closing need to be in your voice. If you’re analyzing data, you might decide the final conclusion must come from your judgment. If you’re writing code, you might decide the architecture decisions are yours to make. Be explicit about this before you start.

WITH AI: YOU DIRECT THE WORK When you’re actually working with AI, think of yourself as a director working with a very fast, very eager, very inexperienced intern. The intern can produce mountains of work quickly, but needs clear direction, makes mistakes, and has no judgment. You wouldn’t turn an intern loose on an important project without supervision. Don’t do it with AI either.

1. Start broad, then narrow. Give AI the task and see what it produces. Don’t expect perfection on the first try. Read the output. Identify what’s working and what isn’t. Then refine: be specific about what to change, what to keep, what to cut, what to expand.

Most good AI work comes from an iterative process, not a single perfect prompt.

2. Ask for alternatives, not one answer. For any significant piece of work, ask AI to produce 2-3

REPLACED 68

After AI: You take ownership

alternatives — different tones, different angles, different structures. This forces you to think about what you actually want, rather than accepting the first plausible answer. Choice is where human judgment enters.

3. Challenge the output. Don’t just accept what AI gives you. Ask it: “What are the weaknesses of this argument?” “What would someone who disagrees say?” “What assumptions am I making here?” “What data would contradict this?” Using AI to argue against itself is one of the most powerful ways to improve your thinking.

4. Push for specificity. AI defaults to generic, bland output. Push it for concrete examples, specific data points, named sources, detailed steps, real numbers. Generic output is not useful. Specific output is.

5. Chain operations. Complex work rarely happens in one prompt. Break work into stages:

research first, then outline, then draft section by section, then edit, then refine. This is analogous to how you’d work with a human assistant — you wouldn’t ask someone to research, write, edit, and finalize a report in one breath. Don’t do it with AI.

AFTER AI: YOU TAKE OWNERSHIP The final phase is where most people fail — and where the value is captured.

REPLACED 69

1. Read the output. Every word. If you’re sending or publishing or presenting AI-assisted work, you must read it completely. Skimming leads to errors, tone-deaf phrasing, hallucinations, and embarrassing mistakes. If you can’t be bothered to read it, don’t use it.

2. Edit in your voice. AI output sounds like AI until you edit it. Rewrite awkward phrases. Add personal touches. Inject a sentence that only you would write. Remove AI clichés (the “in conclusion,”

the “it is important to note,” the bloated hedging). The final output should sound like you, or like the best version of whatever brand or organization you represent.

3. Verify facts and claims. Every number, every citation, every factual claim, every code snippet must be checked if it matters. AI confidently makes things up. It is not lying — it is predicting the most plausible next token — but the effect on your reader is the same. Develop verification habits: cross-check numbers, search for claimed sources, test code, ask for primary sources. If a claim matters and you can’t verify it, remove it or note it as unverified.

4. Add the human element. After editing and verifying, add the one thing AI cannot: your judgment.

Your conclusion. Your recommendation. Your framing of what matters and why. Your sense of what to do REPLACED 70

Chapter 13: Building Your Human Moat

next. This is the layer that turns competent output into work that carries authority.

5. Store and improve your prompts. After each successful use of AI, save the prompt (or template)

that worked, along with notes on what to adjust next time. Over months, you build a personal library of reliable prompts for your recurring work — and that library compounds in value.

The Before-With-After pattern is simple. It takes discipline. It adds a few minutes compared to mindless acceptance of AI output. But the difference in quality between someone who follows this pattern and someone who doesn’t is enormous. One produces augmented professional output that is faster and better than unaided work. The other produces generic, error-prone sludge that makes them look lazy and replaceable.

Chapter 13: Building Your Human Moat In investing, a “moat” is the durable competitive advantage that protects a company from competitors.

In your career, your moat is the set of things about you that cannot be easily replicated — by AI, by a cheaper worker with AI, or by a colleague who learns the same tools.

REPLACED 71

Your human moat is what makes you irreplaceable.

It is what people pay a premium for, even when AI can do a competent version of your job. And here is the good news: you already have the raw material.

Everyone does. The question is whether you deliberately build and deepen it.

Here are the elements of a strong human moat.

1. Specific, lived expertise in a domain that matters. General knowledge is what AI has. Specific, granular, lived experience in a particular industry, company, technology, market, or community is what humans have. If you’ve spent ten years selling industrial equipment to German manufacturing firms, you know things that no AI — trained on generic text — can replicate. You know the buyers. You know the unspoken objections. You know what works in March versus September. You know which plants have which politics. That specific expertise, combined with AI’s general speed, is a powerful moat. Don’t be a generalist in a world of generalist AI. Be deep somewhere.

2. A body of work and reputation. People who have produced a visible body of work — published writing, shipped products, delivered projects, built teams, a track record of results — carry credibility that no AI can match. Your reputation precedes you. It creates trust. It creates opportunities. If you are early in your career, start building that body of work now.

REPLACED 72

Write publicly about your field. Ship things. Deliver results people can point to. A body of work compounds for decades.

3. A network of relationships that trust you. AI can draft an introduction email. It cannot create a twenty-year relationship with someone who will take your call at 7 a.m. because they’ve known you for two decades and you’ve helped them three times. Invest in relationships deliberately, without immediate transactional intent. Be helpful. Be reliable. Show up.

Your network is one of the most moated assets you will ever have.

4. Taste and aesthetic judgment developed over time. As I discussed in Chapter 7, taste is developed by years of consuming and producing the best work in a domain. The person who can look at twenty AI-generated options and instantly identify the best one — who knows why it works, who can refine it into something excellent — has a moat. Taste cannot be automated because it is rooted in thousands of hours of specific experience. Develop taste in your domain: study the best work, understand why it works, build an internal standard of quality.

5. High-stakes judgment under uncertainty. In clear situations with good data, AI can make decent recommendations. In ambiguous situations —

incomplete information, conflicting stakeholders, ethical trade-offs, novel problems with no precedent —

REPLACED 73

AI is useless and can be dangerous. People who have developed judgment through experience, who can make good decisions when the picture is unclear, are extraordinarily valuable. You develop judgment by making decisions, reflecting on them, being wrong, learning, and trying again. Don’t outsource your decision-making. Lean into hard decisions.

6. Character and integrity. This sounds old-

fashioned. It is also the most moated quality of all. A person of known integrity — who tells the truth even when it’s costly, who takes responsibility for failure, who treats people fairly, who can be trusted with sensitive information, who does what they say they will do — builds a reputation that compounds over an entire career. AI has no character. It has no integrity.

It has no loyalty. These qualities will never be commoditized.

7. Communication and presence. The ability to stand in front of a room and move people. The ability to write in a voice that is authentically yours. The ability to listen so deeply that someone feels heard.

The ability to facilitate a meeting where a group of smart people arrive at a decision they all commit to.

The ability to deliver hard feedback without destroying a relationship. These skills are deeply human, and as AI-mediated communication grows, people who can do them in person, powerfully, become more valuable, not less.

REPLACED 74

Chapter 14: Making Yourself Visible in an AI-Augmented Organization

Your moat is not one of these things. It is the combination. A person with deep expertise, a track record, a network, good taste, sound judgment, integrity, and strong communication skills cannot be replaced by AI. They cannot be replaced by a cheaper person using AI. They are the people who use AI to multiply the impact of all those other qualities.

Deepen your moat every year.

Chapter 14: Making Yourself Visible in an AI-Augmented Organization Doing great work quietly is a weak strategy in a time of rapid change. When organizations are restructuring, when teams are being redesigned, when AI is changing how work gets done, the people who get recognized, promoted, and protected are the people whose work is visible — not just their output, but their thinking, their judgment, and their ability to lead in the new way.

Here is how to make yourself visible without being a self-promoter.

Document your AI wins, factually. When you use AI to deliver something faster, better, or with fewer resources, write it down — in a status update, in a weekly email to your manager, in a performance REPLACED 75

review document. Be specific: “I used AI to analyze the customer feedback dataset, which previously took 3 days of analyst time; I completed it in 4 hours and identified three actionable patterns, which led to the change in our onboarding flow last week.” Factual, quantified, tied to business outcomes. This is not bragging. This is accountability and evidence.

Teach what you’re learning. People who teach become the recognized experts. Run a short session for your team on an AI technique that worked for you.

Write an internal wiki page with your best prompts and workflows. Record a five-minute screen share showing how you use AI for a common task. When people in your organization have AI questions, they will come to you. That is visibility, and it is the kind that leads to opportunity.

Speak up about what AI can’t do. As AI enthusiasm sweeps through organizations, there is enormous value in being the person who says, clearly and thoughtfully: “AI can help with this part, but this other part requires human judgment / relationship /

review / risk assessment / customer interaction.”

Leaders value people who can distinguish between where AI creates value and where it creates risk.

Don’t be the AI naysayer. Don’t be the AI zealot. Be the person who understands both.

Own outcomes, not just tasks. In an AI-

augmented workplace, people who can own entire REPLACED 76

outcomes — a project, a client, a product, a metric —

are far more valuable than people who perform individual tasks. Volunteer to own outcomes. Tie your work to business results. When AI helps you complete a task faster, use the time saved to take on more ownership, not just to do more tasks.

Be present. I said it in Chapter 10 and I’ll repeat it: in an increasingly mediated world, physical presence is a differentiator. Attend the meeting. Go to the offsite. Have the in-person conversation. Eat lunch with people. People promote people they know and trust — and trust is built disproportionately in person.

Build a brand outside your organization.

Whether on LinkedIn, a personal blog, industry publications, conference talks, or community leadership, having a reputation outside your company makes you more valuable inside your company. This doesn’t mean becoming an “influencer.” It means having a public body of work and a network in your industry. External visibility gives you internal leverage. It also gives you options.

Visibility is not about performing. It’s about making sure that the people who make decisions about your career — your manager, their manager, the people who run the organization — understand what you contribute and how you’re adapting. In a stable environment, good work eventually speaks for itself.

REPLACED 77

Chapter 15: The Side Hustle, the Portfolio, and the Safety Net

In a fast-changing environment, you can’t wait for it to speak. You have to show your work.

Chapter 15: The Side Hustle, the Portfolio, and the Safety Net One of the most powerful things AI gives you is leverage — the ability to produce more output, test more ideas, and build things faster than was possible even two years ago. That leverage is not just useful in your day job. It is the foundation for a new level of personal optionality.

I am not going to tell you that everyone should quit their job and start an AI-powered business. That is reckless advice, especially if you have dependents, financial obligations, or a job you enjoy. But I am going to tell you that in an era where employment is less stable than it used to be, and where the tools for individual production have never been more powerful, building at least one thing outside your day job is a smart, defensive move.

Here is how to think about it.

Start small, with something you’d do anyway.

Don’t launch a side business that requires forty hours a week. Start with something that is already close to your work or interests: write about what you know, REPLACED 78

build a small tool that solves a problem you have, consult a few hours a month in your area of expertise, create a course or guide for people coming up in your field, sell a template or resource that you wish existed.

AI makes all of these dramatically easier to start.

Use AI to lower the activation energy. The hardest part of building something on the side is the first draft, the first version, the first attempt. AI removes that friction. It can help you write the first draft of a blog post, design the first version of a product, outline a course, build a simple website, draft marketing copy. The first version won’t be perfect, but perfect isn’t the goal — starting is. Use AI to get to version 0.1 fast, then improve it yourself.

Build assets that compound. Some side work pays you once for hours you put in (like hourly consulting). Other work builds assets that can generate value over time — a piece of content that keeps attracting readers, a digital product that keeps selling, a tool that people use, an audience that grows, a brand that opens doors. Prioritize building assets.

One hour creating a durable asset is worth more than five hours trading time for money.

Treat your career as a portfolio, not a single job. The old model was: get a job at a good company, work your way up, retire. That model was already breaking before AI. AI accelerates the break. Think of your career as a portfolio: a primary job (probably), REPLACED 79

some side projects, a network, a set of skills, a reputation, some investments. Each element of the portfolio diversifies your risk and creates new opportunities. If one element fails — if your job is eliminated — you have others to fall back on.

Build in public, at least a little. You don’t need to be a public figure. But sharing what you’re learning and building — even modestly, on a small scale —

creates opportunities. People find you. They offer you work, collaboration, ideas, connections. The person who quietly builds in a vacuum has none of these serendipities. Share what you learn. Put your work out there.

Don’t quit your job prematurely. The safety net is important. A stable income gives you the freedom to experiment on the side without the pressure of needing side income to survive. Most successful side projects take months or years to become meaningful.

Keep your job while you build. When your side income consistently exceeds your salary for six months or more, then consider the leap.

Have a “run-your-own-business” experiment.

Even if you never intend to start a company, spending six months running something small — a newsletter, a shop, a consulting practice, a tool, a community —

teaches you skills that are invaluable in any job:

marketing, sales, customer service, pricing, prioritization, resilience, decision-making under REPLACED 80

uncertainty. Those skills are exactly the skills that AI makes more valuable, because they involve judgment, taste, and human relationships.

The point is not that everyone should become an entrepreneur. The point is that in a world where one person with AI can do what used to take a team, you have the ability to build small bets on the side — bets that can grow into income streams, career pivots, new identities, or just valuable learning. Those bets are your safety net. They are your insurance against disruption. They are also, for many people, where the most interesting work of their lives happens.

REPLACED 81

PART IV: THE LONG GAME

PART IV: THE LONG GAME Chapter 16: Becoming the Person Who Directs the Tool There is a progression that every serious AI user goes through. It is not linear, and it does not happen on a fixed timeline, but it is recognizable. I have seen it in hundreds of people across different fields.

Stage 1: Awareness. You hear about AI. You read articles. You might try a tool once or twice. You don’t yet use it in your work. It feels like a curiosity or a gimmick.

Stage 2: Casual use. You start using AI occasionally — to draft an email, summarize a long document, answer a question. It’s helpful but not transformative. You might use it a couple of times a week.

Stage 3: Regular use. You integrate AI into a few daily tasks. You develop some go-to prompts. You start saving time. You notice it’s changing how you work.

You probably over-trust it sometimes and get burned.

REPLACED 82

Stage 4: Workflow integration. You start building systems — the repeatable workflows I described in Chapter 10. You redesign parts of your work around AI’s strengths and weaknesses. You produce measurably more. You develop a sense of when AI is reliable and when it’s not.

Stage 5: Augmented professional. AI is now a natural part of how you work. You don’t think of “using AI” as a separate activity — you think in terms of getting work done, and AI is one of your primary tools, like a laptop or a phone. Your output is higher quality and higher volume than it was before. You can direct AI on complex tasks because you understand both the tool and your domain.

Stage 6: Director. At this stage — and this is where you want to be — you are not just using AI for your own work. You are directing AI systems, AI-

augmented teams, and AI-integrated workflows. You are making decisions about what AI should do, how work should be divided between humans and machines, where judgment is needed, and how to quality-check the output. You are a leader of augmented work, not just a practitioner of it.

Most people are in Stage 1 or 2. A growing number are in Stage 3. Fewer are in Stage 4. Stage 5 is where the real productivity gains live. Stage 6 is where leadership leverage lives — and it is where the next REPLACED 83

decade of senior roles, promotions, and income growth will concentrate.

The move from Stage 5 to Stage 6 — from individual augmented worker to director of augmented work — is the most important transition of the next five years. Here is what it requires.

You need to understand AI’s limits, not just its capabilities. Most people learn what AI can do.

Directors understand what it can’t do — where it hallucinates, where it’s biased, where it requires human review, what categories of problems it will reliably fail on. This knowledge comes from experience, from study, and from being burned enough times to develop a sixth sense for AI failure modes.

You need to design workflows, not just use tools. Directors don’t just do their own work with AI.

They design how work flows through a team or organization — what goes to AI, what goes to humans, how outputs are reviewed, where feedback loops exist, how quality is maintained, how work is handed off between AI and humans. This is system design, and it’s a new management skill.

You need to develop people, not just tools. In an AI-augmented team, the manager’s job shifts.

Instead of spending time reviewing routine output, you spend time developing people’s judgment, taste, domain expertise, and ability to direct AI well. You REPLACED 84

teach people the verification habits. You help them develop taste. You create an environment where people are not competing with AI but being amplified by it.

You need to make decisions about AI adoption.

When do you adopt a new tool? When do you wait?

Which risks are acceptable? Which processes should be redesigned first? These are judgment calls with organizational consequences, and people who can make them well will run the organizations of the next decade.

You need to retain human authority. The director never lets AI make the final call on anything that matters. The director uses AI for options, analysis, drafts, speed — but the human being retains authority over decisions, especially decisions affecting people, quality, risk, and values.

Becoming a director is not a technical challenge. It is a judgment and leadership challenge. It requires you to be competent with the tools, but far more importantly, it requires you to be thoughtful about how work should happen and how people should be led when the nature of work changes.

That is a role worth aspiring to. And it is a role that AI cannot fill.

REPLACED 85

Chapter 17: Leading Teams in the Augmented Age

Chapter 17: Leading Teams in the Augmented Age If you manage people — a team, a department, a company — AI changes your job in fundamental ways.

Some of the old skills still apply. New ones are required. And many of the default management practices that worked for the last twenty years will be counterproductive in an AI-augmented environment.

Here is what leaders need to understand.

Set the tone on AI adoption. Your team takes its cues from you. If you are visibly learning AI tools, talking honestly about their strengths and weaknesses, and sharing what you’re learning, your team will adopt thoughtfully. If you dismiss AI, your team will either resist (and fall behind) or adopt secretly without guidance. If you hype AI uncritically, your team will over-trust it and make mistakes. The leader’s tone sets the organizational culture around AI. Be the person who models thoughtful adoption:

learning actively, talking openly about failures, setting clear standards for verification, and emphasizing that AI is an amplifier of human judgment, not a replacement.

Measure outcomes, not hours or visible activity. In an AI-augmented workplace, output is no longer tightly coupled to hours worked. A team member who uses AI well might produce in four hours REPLACED 86

what used to take eight. If your management model is built on observing activity — seeing people at their desks, counting hours, tracking keystrokes — you will punish the people who are adapting most effectively and reward those who look busy without producing.

Redesign measurement around outcomes: what was shipped, what problem was solved, what metric improved, what client was served, what decision was made well.

Redesign roles around human-AI collaboration, not around historical job descriptions. Most job descriptions were written before AI changed the economics of cognitive work.

Don’t keep them frozen. Ask: for each role on your team, what work should be done by AI? What work should be done by a human with AI assistance? What work must be done purely by a human? Redesign roles accordingly. This doesn’t mean cutting headcount blindly — it means freeing people from the Category 1

routine work so they can spend more time on Category 2 and 3 work, where human judgment and relationships create value.

Invest in verification skills and quality standards. The biggest risk of team-level AI adoption is a collapse in quality control. If everyone on a team is using AI and nobody has rigorous verification habits, errors propagate. Hallucinations get copied into documents. Generic AI-speak ends up in client communications. Mistakes compound. Set explicit REPLACED 87

quality standards. Build verification into your workflow — peer review, spot checks, clear sign-off requirements. Make it a point of team culture that no AI output goes to a client, a senior leader, or the public without a human being taking full ownership.

Make decisions about AI access and equity. Not everyone on your team will adopt AI at the same pace.

Some will be early adopters; some will resist. This creates an internal productivity gap that can become political and unfair if unmanaged. Provide training.

Set baseline expectations for AI literacy. Don’t let the early adopters capture all the gains while the resisters get left behind. But also don’t force adoption so aggressively that people use AI poorly or inappropriately. Set standards. Provide support. Give it time.

Protect time for thinking and human work. AI makes it easy to fill every minute with output — more drafts, more emails, more meetings, more reports. But the most valuable work of leadership and knowledge work often happens in unstructured time: thinking, reading, reflecting, walking, talking to people one-on-

one, building relationships, staring at a problem. Don’t let AI-enabled productivity fill all available space.

Protect white space on your calendar and on your team’s calendars. Value thinking, not just output.

Lead with courage on the hard questions. AI will raise questions that do not have easy answers:

REPLACED 88

Which roles should be reduced? Which new roles should be created? How do you treat people whose roles are changing? How do you handle layoffs humanely if they are necessary? How do you ensure AI is used ethically? How do you protect customer and employee privacy? These are not technical questions — they are leadership questions. They require judgment, empathy, and moral courage. Avoiding them doesn’t make them go away; it makes the eventual transition worse.

Model the behavior you want. If you as a leader use AI thoughtfully — verifying outputs, maintaining your own thinking, being transparent about when you use AI, taking responsibility for AI-assisted mistakes, crediting AI where appropriate, not pretending you wrote something AI wrote — your team will follow your example. If you use AI carelessly, pass off AI writing as your own, or expect your team to do things you won’t do yourself, you will erode trust and create a culture of corner-cutting.

Leadership in the augmented age is not about becoming a technical AI expert. It is about being a thoughtful human being who understands what the technology can do, what it cannot, and how to organize people to use it well while preserving their humanity, their development, and the quality of their work.

REPLACED 89

Chapter 18: Staying Human When the Machines Get Better Every Year

Chapter 18: Staying Human When the Machines Get Better Every Year Let me end the main section of this book with the question I think about more than any other.

AI will improve. Every year, the models will get better. They will handle more tasks more reliably. They will make fewer errors. They will write more fluidly, code more correctly, analyze more accurately, reason more capably. Every year, the bar for what constitutes “human-only” work will rise a little higher. Some tasks that I have told you in this book are safe will become partially automatable. Some skills that feel irreplaceable today will be partially replicable tomorrow.

If you are playing a static game — “I will do X, which AI can’t do, forever” — you will eventually lose.

The machines keep getting better.

So the long game is not to identify a fixed island of human-only work and defend it. The long game is to stay human while using every tool available — to develop the qualities that are not just currently un-

automatable but are inherently human, qualities that become more valuable the more powerful machines become.

REPLACED 90

What are those qualities?

Curiosity. The desire to know, to understand, to ask questions no one else is asking. AI can answer questions. It doesn’t genuinely wonder. Cultivate curiosity as a practice: read widely, ask why, follow threads that interest you even when they’re not “productive,” remain fascinated by the world. Curious people adapt because they are always learning.

Courage. The willingness to take risks, to be wrong, to try things that might fail, to say what you believe even when it’s unpopular, to make decisions without complete information. AI has no courage. It cannot take a genuine risk because it has nothing at stake. Courage is human. In a world where safe, conventional answers can be AI-generated, the person who takes intelligent risks — who launches the thing, who says the uncomfortable thing, who commits to the uncertain path — stands out.

Care. Genuinely caring about other people, about the quality of work, about the impact of what you do, about customers, about colleagues, about the community you operate in. AI can simulate care. It cannot feel it. People can tell the difference over time.

The people who genuinely care — about their craft, about their team, about their customers — build loyalty and trust that no simulation can replicate.

Presence. Being fully where you are. Listening without distraction. Attending to the person in front of REPLACED 91

you. Noticing what’s happening in a room, in a moment, in a conversation. In a world of infinite distraction and AI-mediated communication, genuine presence is rare and powerful.

Creativity at the root. I don’t mean remixing existing ideas — AI can do that. I mean the kind of creativity that comes from a deep engagement with life: from your unique experience, your suffering, your joy, your relationships, your observation of the world from the specific vantage point that only you occupy.

The greatest art, the most original ideas, the most transformative businesses, come from people who have lived deeply and brought something of their specific humanity into their work. That is not automatable because no machine has lived your life.

Moral conviction. Knowing what you stand for, what you will not do, where the lines are. AI has no values. It has no conscience. It operates within the constraints humans give it, but it does not genuinely care about right and wrong. People who have clear values, who will make personal sacrifices for what they believe is right, who will refuse work that violates their conscience, are the people who provide the moral compass in organizations and in society. That becomes more important, not less, as technology grows more powerful.

Love and connection. This is the deepest human moat of all. The love you have for people — your REPLACED 92

family, your friends, your colleagues, your community — is what gives work meaning. It is what you are ultimately working for. It is what makes you a human being rather than a productivity engine. Never let the tools of productivity cause you to neglect the relationships that give your life purpose. That way lies emptiness, no matter how productive you become.

The future belongs to people who are more human, not less, as machines get more powerful. It belongs to people who use AI to free up time and attention for the things that are most human: thinking deeply, creating originally, caring deeply, leading courageously, loving fully, living deliberately.

The alternative is a world where people act like machines — following AI prompts, producing generic output, optimizing for metrics, outsourcing their thinking — and that is a world where you become replaceable, no matter how well you use the tools.

Don’t become a machine to compete with machines.

Become more fully human, and use machines as extensions of your humanity. That is the ultimate competitive advantage. It is also the only approach that leads to a life and a career worth having.

REPLACED 93

Afterword: The Choice in Front of You

I Afterword: The Choice in Front of You want to leave you with a simple thought.

You are not powerless in this transition.

You are not a victim of technological change. You are not fated to be replaced, nor guaranteed to thrive.

You have agency. You have choices. The choices you make in the next year — about how you work, what you learn, how you spend your time, which skills you develop, how you relate to the people around you —

will determine where you end up.

The choice is not “embrace AI or reject AI.” That is a false choice. The technology exists. It is being deployed. It will be used by your competitors, your colleagues, your employer, and your industry. You cannot opt out of a global technology shift, any more than you could have opted out of email or the internet.

The real choice is this:

Will you use this technology thoughtfully, deliberately, and in service of your own humanity — to do better work, free up time for what matters, develop yourself, serve your clients or customers, and build a career that compounds?

REPLACED 94

Or will you drift — using AI half-heartedly, outsourcing your thinking, neglecting your human skills, hoping for the best, until one day you look up and realize the world has moved around you?

The first path requires effort. It requires practice. It requires honesty about what you don’t know. It requires you to be proactive rather than reactive. It requires you to stay human even when the machines are doing more of the work.

But it is also the path that leads somewhere worth going. It leads to being the person who uses the tools — not the person the tools use. It leads to being the one who is irreplaceable, not because you refuse to change, but because you have combined the power of the technology with the depth of your own humanity.

That is the person who thrives. That is the person who leads. That is the person who is not replaced.

You can be that person. The window is open. The tools are available. The evidence is clear.

Start today.

REPLACED 95

Appendix: The Evidence — Studies Referenced in This Book

T Appendix: The Evidence — Studies Referenced in This Book his appendix summarizes the key research studies cited throughout the book, with enough detail that you can verify or follow up on any claim that matters to you. Where studies have been published in peer-reviewed journals, I have noted the journal. Where they are working papers, corporate reports, or organization analyses, I have noted that as well. This is not an exhaustive survey of the AI productivity literature — it is a curated selection of REPLACED 96

the most rigorous and informative studies available as of late 2026.

Brynjolfsson, E., Li, D., & Raymond, L. R.

(2023/2025). “Generative AI at Work.”

NBER Working Paper No. 31161 (first circulated 2023, revised 2025). Studied 5,179

customer support agents at a Fortune 500

software firm using a generative AI assistant in a staggered rollout. Found 14% average increase in issues resolved per hour; 34%

increase for novice agents (under 2 months tenure); improvements in customer sentiment and reduced agent attrition; gains largest for least-skilled workers.

Noy, S., & Zhang, W. (2023).

“Experimental evidence on the productivity effects of generative artificial intelligence.”

Science, 381(6654), 187-192. Preregistered online experiment with 453 college-educated professionals across writing, analytical, and marketing tasks. ChatGPT access produced 40%

reduction in time spent and 18% improvement in quality (as graded by blind evaluators).

Effects concentrated among workers with weaker writing skills.

1.

2.

REPLACED 97

Dell’Acqua, F., et al. (2023/2024).

“Navigating the Jagged Technological Frontier: Field Experimental Evidence of the Effects of AI on Knowledge Worker Productivity and Quality.” Harvard Business School Working Paper 24-013. Studied 756 BCG consultants across 18 tasks. On tasks within AI capability, consultants completed 12.2% more tasks, finished 25.1% faster, and produced 40%

higher-quality outputs. On a task outside the AI’s frontier, AI-using consultants were 19

percentage points less likely to produce correct solutions due to over-reliance on incorrect AI output.

Peng, S., Kalliamvakou, E., Cihon, P., &

Demirer, M. (2023). “The Impact of AI on Developer Productivity: Evidence from GitHub Copilot.” Cornell/arXiv working paper.

Randomized controlled trial with 95 developers;

treatment group completed tasks 55.8% faster.

Cui, Demirer, Jaffe, Musolff, Peng & Salz (Management Science, 2026) extended to 4,867

developers, finding 26% more tasks completed with GitHub Copilot, with largest gains for less-

experienced developers and those who adopted deeply. The METR (Ecotrust) replication study of late 2024 found experienced developers initially 19% slower, a result that did not hold in 2025 replications as models and user fluency improved.

3.

4.

REPLACED 98

PwC Global AI Jobs Barometer (2025).

Analysis of labor market data across major economies. Found that workers with AI skills are earning a wage premium of up to 56% over comparable workers without AI skills; the premium roughly doubled over the preceding 12 months. Sectors most exposed to AI are seeing both faster employment growth and faster productivity growth.

EY Work Reimagined Survey (2025).

Survey of thousands of workers and employers across 25 countries. Found 88% of employees report using AI at work, but only 5% use it in advanced or transformative ways integrated into workflows; 37-43% of workers express concern about skill erosion from over-reliance on AI.

World Economic Forum, Future of Jobs Report (2025). Projections covering 1,000+

employers representing 65 million workers globally: 92 million roles displaced and 170

million new roles created by 2030 (net +78

million); 39% of core skills will change; 59% of the global workforce will require retraining. AI and machine learning specialists top the list of fastest-growing roles.

5.

6.

7.

REPLACED 99

Yale Budget Lab + Brookings Institution (October 2025). Joint analysis of U.S.

employment data since the launch of ChatGPT (November 2022). Found no statistically significant aggregate employment impact attributable to AI to date. Challenger, Gray &

Christmas data showed 17,375 job cuts explicitly attributed to AI through September 2025, against roughly 1.17 million total layoffs in the same period.

World Bank, Africa’s Pulse + Africa AI report (2025). Analysis of AI adoption in African firms. 74% of firms expect AI to augment rather than replace workers; youth entry into the labor market increasingly requires “AI operator” skills; African firms lag in AI adoption but are catching up in mobile-first sectors.

Detecon/Rahim Hirji Meta-Analysis (2026). Review of 555 graded studies on AI and work performance. Found automation bias (over-reliance on automated systems) is robust across aviation, medicine, military, and knowledge-work domains; AI error rates are non-trivial on out-of-distribution tasks; skill fade from over-automation documented across multiple domains.

8.

9.

10.

REPLACED 100

MIT Media Lab (2025). Experimental study finding cognitive costs of AI reliance on creative and analytical writing tasks: participants who repeatedly used AI for ideation and drafting showed reduced performance on subsequent unaided tasks, suggesting a form of cognitive atrophy from sustained over-reliance.

Acemoglu, D. (2024/2025). Multiple papers, lectures and NBER work on the macroeconomics of AI. Argues near-term productivity gains from AI are likely to be modest (well below historical transformative technology levels) due to task limits, organizational complementarities, and the need for significant complementary investment.

Emphasizes that technology’s impact depends on how it is deployed and governed, not on the technology itself.

Eloundou, T., Manning, S., Mishkin, P., &

Rock, D. (2023). “GPTs are GPTs: An Early Look at the Labor Market Impact Potential of LLMs.” OpenAI working paper. Estimated that roughly 80% of the U.S. workforce could have at least 10% of their work tasks affected by LLM introduction; 19% of workers may see at least 50% of their tasks impacted. Widely cited as a starting framework for exposure analysis, though criticized for measuring task exposure rather than job displacement.

11.

12.

13.

REPLACED 101

Notes

Dallas Federal Reserve (2025). Analysis finding that workers in AI-exposed industries who have AI skills are seeing wage growth, while those without AI skills in the same industries see wage stagnation.

Goldman Sachs Global Economics (2023/2024). Estimated that generative AI could raise global GDP by 7% over a decade and drive labor productivity growth of 1.5

percentage points per year, with roughly two-

thirds of occupations exposed to some automation, but historical precedent suggesting new job creation offsets displacement over time.

A note on the research: The field is moving fast.

Studies published in 2023 used GPT-3.5 and early GPT-4; later studies use substantially more capable models. Findings will continue to evolve. I have tried to avoid cherry-picking and to note where results are contested or where replication has produced conflicting findings. As you read and apply the ideas in this book, I encourage you to follow the research yourself — the story is far from over.

Notes The productivity statistics cited in this book are drawn from the studies listed in the appendix. Where I have 14.

15.

REPLACED 102

Acknowledgments

given specific numbers (14%, 34%, 40%, 26%, 56%), these correspond to point estimates from individual studies and should be understood as approximate ranges — effects vary by setting, worker skill level, task type, and model version. They are intended to illustrate magnitudes, not to predict a precise productivity gain for any individual worker.

The frameworks in this book (AUGMENT, the three-

category skill framework, Before/With/After AI, the six stages of AI adoption) are syntheses developed from conversations with practitioners, study of the research literature, and my own observation and experience.

They are not empirically validated instruments — they are practical tools designed to help you think and act.

The advice in this book is general. Your situation is specific. Adapt the frameworks to your role, your industry, your organization, and your goals. There is no one right way to work with AI — there is only the way that produces better outcomes for you, consistent with your values and the quality standards you set for yourself.

Acknowledgments This book would not exist without the researchers whose work I have drawn on throughout —

particularly Erik Brynjolfsson, Danielle Li, Lindsey REPLACED 103

Raymond, Shakked Noy, Whitney Zhang, Fabrizio Dell’Acqua, Daron Acemoglu, Sida Peng, and their many co-authors, whose careful empirical work has given us an evidence base rather than slogans and speculation. Any errors or misinterpretations of their work are mine, not theirs.

I am grateful to the many professionals, founders, managers, and workers who spoke with me about their experiences adapting to AI in their own work —

named and unnamed. Your stories are the spine of this book.

I wrote this book in English for a global audience, but I wrote it as someone who comes from a part of the world where every technological shift has been both a threat and an opportunity. I hope this book helps people everywhere — in Harare and Helsinki, Lagos and London, Nairobi and New York — to navigate this shift with agency and confidence.

Most of all, to the reader: thank you for picking up this book. I hope it helps you. The future is not something that happens to you. It is something you help build, one decision, one skill, one day at a time.

— Courage Garikai REPLACED 104