Life after the Death of a Middleman

Redesigning the blueprint for a resilient life in the age of AI

If the middleman is sometimes us, how should we build a life around that possibility?

That was the question left hanging at the end of my article:

The title of this article is a nod to Arthur Miller’s Death of a Salesman. Miller’s subject was very different, but his play raised questions about work, success and what happens when the expectations on which we build a life no longer hold up quite as well as we thought. Those questions seem surprisingly relevant here.

After tracing disintermediation from banking through travel, retail and investing, I argued that artificial intelligence was extending the same process into expertise, while AI combined with robotics could eventually broaden its reach into physical work. The uncomfortable conclusion was that many of us might occupy the position that technology is increasingly capable of bypassing.

The obvious next question: What should do about it?

The obvious answer is to predict what happens next. Which professions are safe? Which skills should we learn? How quickly will robotics develop? What will happen to salaries? Which jobs will disappear, and what new ones will take their place?

I don’t think we can answer those questions with much confidence. AI is already changing the nature of work, but even organizations studying its impact emphasize that it may complement people in some tasks, replace them in others and create new kinds of work. The eventual balance remains uncertain.

Fortunately, we don’t need to predict the future correctly to recognize the problem. If the world of work becomes less predictable, then some of the assumptions on which we have built our lives become less dependable too. And that suggests a different question. Rather than asking how to predict the future, we should ask how to build a life that doesn’t depend too heavily on following one well planned out, predetermined path.

The Blueprint Was Built on Assumptions

I’ve written before in Actualize This about the traditional blueprint for a successful life. Study hard. Acquire valuable skills. Get a good job. Build a career. Buy a home. Raise a family. Save. Retire.

There is nothing inherently foolish about this blueprint. Quite the opposite. For millions of people it provided a sensible way to organize the financial and practical stages of adult life. Education created knowledge and credentials. Those credentials opened the door to a profession. Experience increased expertise. Expertise increased economic value. Greater economic value produced higher income, wealth and greater financial security. On that foundation, people could build the rest of their lives.

But note how many assumptions are hiding inside that sequence and how many of those long-held assumptions are currently are not holding so well. Education has to lead to capabilities that remain valuable. Employers have to continue needing those capabilities. Expertise has to remain sufficiently scarce to command an attractive salary. A career has to remain reasonably durable for long enough to support mortgages, children, retirement savings and all the other commitments accumulated along the way.

AI and robotics do not have to destroy that chain to become one of our overriding concerns for our futures. They merely have to weaken some of its links. A qualification can remain valuable without guaranteeing that the expertise behind it remains scarce for an entire career. A respected profession can remain important while some of its economically valuable tasks become easier to automate or perform directly. A skill can remain useful while depreciating faster than it once did. Previous era defining technologies have redefined professions. AI’s impact is not a new phenomenon, but the speed an breath of its impact are expected be faster and broader than any preceding technological revolution.

This matters because the traditional blueprint concentrated quite a lot of risk in one place: our careers. For most of us, this hardly felt like risk. A career was the opposite. It was the mechanism through which uncertainty diminished. You accumulated experience, moved upwards, earned more and gradually transformed your human capital into financial capital.

That model certainly hasn’t disappeared. But if the economic value of particular tasks, skills and forms of expertise can change more quickly, then following the blueprint faithfully may no longer produce quite the same confidence about the destination. A good blueprint doesn’t have to be torn up when conditions change. But it does have to be redesigned to remain effective in planning for success.

Careers or Capabilities?

This raises a problem in the way we often think about human capital. We tend to organize it around careers. We choose an area of study, acquire a qualification, enter a profession, accumulate expertise and progress through some version of an occupational hierarchy. The career becomes the durable asset, while the capabilities accumulated along the way are treated almost as components of that career. Perhaps that relationship needs reversing.

AI and The Great Disintermediation argued that jobs are the wrong level at which to think about technological disruption. A lawyer, accountant, analyst, teacher or engineer does not perform one indivisible activity. Each occupation consists of many tasks. Technology can reduce the value of some while leaving others relatively untouched, at least for a time. AI may perform preliminary research without accepting professional responsibility. It may produce a first draft without making the final judgment. Robotics may perform one physical operation without possessing all the adaptability required for the occupation surrounding it.

The International Labour Organization has similarly found in its work on generative AI that augmentation of tasks may be more important than wholesale automation of occupations, although exposure varies substantially between types of work and economies.

This suggests a subtle change in the question we ask ourselves. Instead of focusing entirely on “What career should I choose?”, perhaps we should pay more attention to “What capabilities should I keep developing?”

There is nothing revolutionary about lifelong learning, transferable skills or skills-based (rather than knowledge-based) education. People have been talking about all three for decades, often while attending corporate training sessions on PowerPoint. But AI potentially increases their importance because the useful life of particular forms of expertise may become less predictable.

The OECD argues that adapting to AI-related labor-market change will require developing new skills as demands evolve, while its work on AI and skills emphasizes learning throughout working lives.

What matters may therefore be less a particular inventory of supposedly future-proof skills and more the ability to continue acquiring and combining capabilities. Substantive knowledge still matters. So do judgment, creativity, critical analysis, technological fluency and the ability to distinguish reliable information from material that is synthetic, mistaken or misleading. But I would be wary of declaring any of these permanently protected from technology.

Also, merely having this list of skills is not a recipe for success for 2026 and beyond. It is the capacity to keep adapting when the list changes. Human capital starts to look less like a single asset purchased through education and more like a portfolio requiring continuing investment. Some capabilities appreciate. Others lose value. New ones become important. Existing expertise can be combined with new technology rather than simply defended against it.

Perhaps the more useful question is no longer, “What career will insulate me from AI transformation?”, but “How do I remain capable when the value of particular tasks changes?”

When Security Moves Back Up the Hierarchy

This brings us back to money. In The Hierarchy of Money, I argued that money matters differently depending upon what we need it to do. At the lower levels of the hierarchy its importance is difficult to overstate. Money pays for food, housing, healthcare and the basic infrastructure of everyday life. Beyond that, financial resources provide security, protection against shocks and freedom from the anxiety of constantly worrying about whether the next bill can be paid.

As those needs become securely satisfied, however, the marginal value of additional money can diminish. More income can still be useful, but relationships, autonomy, meaning, growth and other elements of flourishing may become increasingly important. The point isn’t that money stops mattering. It is that what we need from money changes.

There is an important dynamic implication here that is easy to miss: you don’t permanently graduate from security. Imagine someone with a stable career, predictable income, manageable mortgage and adequate savings. This person may quite reasonably decide that the next salary increase matters less than having more time with family, doing more meaningful work or pursuing interests outside the office. The financial foundations feel sufficiently secure that attention can move elsewhere.

Now introduce a credible threat to that income. Nothing fundamental about money has changed, but its importance probably has. Financial security moves back toward the foreground because the need it satisfies has become less secure.

Research on household financial shocks supports the basic intuition that buffers matter. One U.S. study of lower- and moderate-income households found that income shocks were associated with declines in subjective financial wellbeing, with relatively liquidity-constrained households tending to be more negatively affected. (Financial Shocks and Financial Well-Being)

The point here isn’t to generalize from one group to everyone or prescribe a particular level of savings. It is that resilience has value precisely because we don’t know when it will be needed. If employment becomes less predictable, buffers, manageable commitments, flexibility and adaptable earning capacity all become more valuable. A financial reserve isn’t merely a pile of money earning interest. It can buy time to retrain, refuse the wrong opportunity, tolerate a period of lower income or make a deliberate change rather than an emergency one.

This is why increased technological uncertainty doesn’t make the Hierarchy of Money less relevant. It makes its dynamic nature more obvious. Financial security creates the room from which we can pay attention to things beyond financial security. If that foundation becomes less stable, money naturally demands more of our attention again. The objective, therefore, isn’t maximum wealth. It is sufficient security to preserve options when circumstances change.

The AI Paradox

So far, we have considered AI principally as a potential source of economic uncertainty. But that is only half the picture. On one side is work. If AI reduces the scarcity value of tasks that currently support our income, career paths can become less predictable and financial resilience more important. The eventual effect remains uncertain.

The OECD expects AI to complement people in some tasks, replace them in others and generate new kinds of work, while the International Labour Organization notes that translating task-level productivity gains into wider economic gains depends on factors including skills, workplace reorganization and complementary investment.

But what about the other side of the equation? A great deal of knowledge used to be expensive to obtain. Tuition cost money. Specialist advice cost money. Access to sophisticated tools cost money. Learning a subject often required buying books, attending courses or finding a person prepared to teach you. None of those barriers has disappeared, but AI can reduce some of them dramatically.

A person trying to understand a difficult subject can ask for an explanation at almost any level of complexity. Someone learning a new skill can receive examples, feedback and practice material. A person developing a creative project can experiment in ways that previously required more time, money or specialist assistance. In AI and the Great Disintermediation, I described this as the other side of AI-driven disintermediation: inexpensive access to portions of expertise previously concentrated among professionals. Here the consequence becomes broader. Some of the resources we use to develop ourselves may also become cheaper and more accessible.

That gives us the AI paradox: AI may make financial security harder to achieve for some people while simultaneously making some of the tools for personal growth easier to access. Those effects are not mutually contradictory because financial security and flourishing are not the same thing. Financial security can provide an essential foundation for a good life without guaranteeing that the life built upon it will be good. Likewise, access to inexpensive knowledge and powerful tools does not remove the need for security.

The same technology can therefore push different parts of wellbeing in different directions. It can introduce uncertainty while creating opportunity. It can reduce the economic scarcity of something we sell while reducing the financial cost of something we want to acquire. That is why describing AI as either a great threat or a great democratizing force seems inadequate. It can plausibly be both.

What Can’t We Outsource?

This paradox however comes with a caveat. AI may give us cheaper access to knowledge, advice, coaching, explanation and creative assistance. But access to the tools for living well is not the same thing as actually living well.

The distinction should be familiar from money. Money can improve the conditions in which a good life becomes possible. It can provide housing, safety, education, healthcare, experiences, freedom and time. But accumulating money isn’t the same activity as building relationships, developing competence, finding meaning or flourishing. AI seems capable of creating a similar confusion because its outputs can feel unusually close to the desired outcome.

An AI system can explain how exercise works, help construct a training program and answer questions about technique. None of those things makes me fitter while I sit on a sofa reading the answers. AI can suggest ways of dealing with a difficult conversation, but it cannot have that conversation for me in any meaningful sense. It can teach, explain, quiz and correct, but there remains a difference between having an excellent tutor available and actually acquiring competence.

The same applies to creativity and purpose. AI can help explore an idea, challenge assumptions or make it easier to turn an intention into something tangible. Those capabilities can be extraordinarily useful. But easier access to creative tools doesn’t guarantee that someone will create anything they find meaningful. Nor does another machine’s ability to discuss purpose relieve us of the rather inconvenient requirement to decide what matters to us.

I don’t think we need to turn this into a grand claim about uniquely human abilities that machines can never reproduce. History offers plenty of reasons to be suspicious of confident statements about technological boundaries. The more durable distinction is between tools and lived outcomes. Some things matter partly because we participate in creating them. Competence is developed. Relationships are lived. Health is maintained. Character is formed through choices and behavior. Meaning has to belong to the life being lived.

Just as money can provide the conditions for flourishing without purchasing it, AI may provide increasingly powerful tools for growth without doing our growing for us. That brings us to the question I find myself thinking about most.

What Should We Teach Our Children?

When I was learning about disintermediation at business school thirty years ago, the conventional advice about building a successful working life seemed reasonably straightforward. Study hard. Get good qualifications. Acquire valuable expertise. Work hard. Build a career. Buy a home. Save. Eventually retire. Much of that still sounds like good advice. What I’m less certain about is whether it is enough.

The question looks different to me now because I’m no longer the student receiving the advice. I’m a parent wondering what guidance makes sense for children whose working lives will stretch decades into a future I cannot predict. I don’t know which occupations will be most valuable, which technologies will mature, what jobs will emerge or how many times they may need to change direction. Telling them to choose the correct career would therefore require me to possess precisely the forecasting ability I’ve spent this article arguing we don’t have.

That doesn’t mean education no longer matters, that expertise is pointless or that careers are about to disappear. It means we should be careful about treating any one of them as a permanent guarantee. These things may still matter. They may simply no longer be sufficient.

There is another reason this question is useful. Asking what we should teach our children creates a convenient illusion that the problem belongs to them. It doesn’t. AI is already entering today’s workplaces.

The OECD describes AI as increasingly changing the tasks performed in jobs and emphasizes learning throughout working lives, while its 2026 Skills Summit placed adult skills and opportunities across generations alongside the need to rethink formal education.

Someone beginning work today may face these changes for decades, but someone halfway through a career may still have twenty years of work ahead. Even those approaching retirement may discover how much of their routine, identity and social world has been supplied by work. So perhaps the question is broader: what would we want our children to build if we could not predict the working world they are entering?

I can think of at least five places to build resilience. The first is financial security. Not maximum wealth and certainly not money as the objective of life, but enough resilience to absorb disruption without every setback becoming an immediate threat to basic security. Savings, manageable commitments and financial flexibility can create something particularly valuable in an uncertain world: being able to buy time to respond.

The second is adaptable capability. I would still want my children to become good at things. Expertise matters. Qualifications can matter. Hard work matters. But I would be more hesitant to assume that one qualification, obtained at 22, should define the economic basis of the next forty years. The deeper asset is the ability to continue learning, exercising judgment, evaluating information, using new tools and combining old expertise with new capabilities.

The third is health. This can sound strangely obvious until we notice how often the traditional blueprint encourages us to treat health as something to optimize after the career is established and the finances are sorted out. Physical and mental capacity affect our ability to adapt, work, maintain relationships and make use of opportunities. A more resilient blueprint treats health as infrastructure rather than a future project.

The fourth is relationships. Work can provide valuable friendships, community and human connection, but it is risky when too much of our social world belongs to our professional world.

The World Health Organization notes that decent work can provide not only a livelihood but also confidence, purpose, achievement, positive relationships, community inclusion and structured routines. That is precisely why losing or changing work can affect far more than income.

The fifth is purpose beyond occupation. I hope my children find work they consider meaningful. But I am less sure I would want salary, job title, promotion and professional status to become the principal scorecard through which they determine whether their lives are going well. If work is the sole source of identity and meaning, disruption to a career can become disruption to the self.

These five foundations are not a new theory of wellbeing, nor are they an exhaustive list. Similar themes appear throughout research on wellbeing and flourishing. I am using them here for a narrower purpose: to think about how to build a life that is less dependent on the durability of one career. They also cannot simply substitute for one another. Money cannot fully compensate for poor health. Health cannot pay the mortgage. Adaptability cannot replace close relationships. Purpose does not eliminate the need for financial security.

That is the point. Resilience comes partly from not asking one thing to do everything.