There are three phases of human economic development. We have lived through two of them. We are, right now, in the early weeks of the third.
The first phase lasted roughly four hundred years, from the first mechanised looms through to the microchip. Call it the era of diminishing returns. Growth was linear, effort-intensive, and bounded by the limits of what human hands and mechanical systems could produce. Every doubling of output required a near-doubling of input — more people, more capital, more hours, more physical plant. The candlelight era, if you wanted an image for it. Slow, patient, incremental, and plateau-prone. Most of human history fits in this phase, and most of our instincts about how work, business, and economic growth actually behave were formed inside it.
The second phase began in the 1960s with the integrated circuit and accelerated through the 1990s with the networked computer. Call it the era of exponential growth. Gordon Moore noticed that transistor density on a chip doubled roughly every two years, and that observation — Moore's Law — turned out to describe not just semiconductors but the entire digital economy. A generation of entrepreneurs, investors, and policymakers learned the strange mathematics of exponential curves: quantities that look almost stationary for a decade and then become dominant in five years. Punch cards gave way to personal computers, which gave way to smartphones, which gave way to networked software eating every industry it touched. Roughly a billion doublings over ninety-nine years. Exponential growth is dizzying if you're used to linear, but it has a quality that makes it still human-legible: you can feel the curve bending, even if you can't quite keep up.
Phase Three is not more of Phase Two. That is the single most important sentence in this essay, and the one most commercial conversations about AI get wrong.
Why Phase Three is qualitatively different
In Phase Two, growth was exponential but the unit of production was still largely human. A software engineer wrote the code. A designer made the interface. A marketer wrote the copy. Technology amplified human effort — dramatically — but the effort itself remained human, and the rate of growth was bounded by how quickly humans could learn, hire, and scale.
In Phase Three, that boundary begins to dissolve. The defining feature of Phase Three is recursion: systems that produce outputs — including systems that produce better systems — with progressively less direct human input. This is what makes the growth curve bend differently. Not faster exponential. Hyperbolic.
The mathematical distinction matters. An exponential curve has a constant doubling time. A hyperbolic curve has a doubling time that itself shrinks. In practice, that means Phase Two doubled capability roughly every two years. Phase Three is doubling capability, in some domains, every few months — and the doubling interval is itself getting shorter. If Moore's Law gave us a billion doublings over ninety-nine years, the early trajectory of Phase Three suggests the equivalent compounded capability in under six years. That is not a forecast. That is the arithmetic of the curves as we are currently measuring them.
Moore's Law, the engine of the last forty years, now looks like a leisurely Sunday driver being overtaken by an F-18 taking off.
The image matters because the feeling of Phase Three is different from the feeling of Phase Two. Exponential growth was thrilling. Hyperbolic growth is vertiginous. The rules that made a business defensible in Phase Two — scale, brand, network effects, proprietary data — are not necessarily the rules that make a business defensible in Phase Three. They may be. They also may not be. The category is unsettled in a way that Phase Two, for all its speed, never was.
The two-to-three year window
Human systems adapt slowly. Humanity had roughly a hundred years to absorb the transition from Phase One to Phase Two, and we barely managed. The social, political, and organisational upheavals of the twentieth century are, in large part, the sound of institutions trying to catch up with an economy that had fundamentally changed shape.
The transition from Phase Two to Phase Three does not grant us a century. On current trajectories, the critical adaptation window is two to three years. Not the full transformation — that will play out over a decade or more. But the decisions that determine whether your organisation is on the right side of the transition will, for most businesses reading this, be made between now and 2028. After that, the organisational, cultural, and market positions will largely have hardened, and the cost of getting them wrong will have compounded past the point of recovery.
This is not a call to panic. Panic is a Phase Two response — norepinephrine, deadline-driven, exhausting, and almost always wrong. It is a call to attention. The leaders who navigate Phase Three well will be the ones who have noticed what phase they are in, and who are building businesses, strategies, and themselves for its actual mechanics rather than for the residual instincts of the phase just past.
What this means for leaders
The most common mistake we see leaders making is treating Phase Three as a technology problem. It isn't. The technology is the mechanism. The problem is everything downstream of the mechanism: how you organise work, how you price outputs, how you compete, what skills you hire for, what your business model actually is when the cost of cognitive work approaches the cost of the electricity running it.
The second-most-common mistake is assuming that the leaders who thrived in Phase Two will, by default, thrive in Phase Three. Most won't. The cognitive patterns that made a Phase Two leader successful — rapid decision-making inside a known frame, relentless execution of a proven playbook, comfort with exponential but not hyperbolic change — turn out to be actively maladaptive in a phase where the frame itself is being rewritten faster than the playbook can be executed.
Phase Three leadership is a different set of capacities. The ability to hold multiple possible futures in mind without prematurely collapsing to one. Comfort sitting with ambiguity for longer than Phase Two instincts want to tolerate. A willingness to rebuild the playbook mid-quarter rather than defend the one that is already failing. And — this is where most transformation work stops short — an inner capacity that our work has historically framed as spiritual intelligence: the ability to stay steady when the external map is being redrawn weekly. Hyperbolic environments reward inner stability as strongly as they reward outer agility.
This is the part of Phase Three that AI will not solve for you. AI is the mechanism producing the hyperbolic curve. Your thinking is what decides whether that curve lifts you, flattens you, or passes you by. That is the leader upgrade Phase Three demands, and it is the upgrade that our work — in the Future Business Growth Model, in our Pods, in our Briefings — is most directly designed to support.
Where the Law of Diminishing Effort leads
The Law of Diminishing Effort is not a metaphor or a brand story. It is a description of an economic phase that began, by our measure, around 2023 and will define the commercial landscape for at least the next decade. The framework is testable. The curves are measurable. The business consequences are already visible in the firms moving fastest.
What it does, done well, is give leaders a sharper vocabulary than the generic "AI transformation" language allows. It names the phase. It distinguishes hyperbolic growth from its exponential predecessor. It clarifies why the adaptation window is short and why the leader upgrade is non-negotiable. And it points — via the Future Business Growth Model — at the organisational architecture that Phase Three businesses will need.
Most of our clients arrive with a strong intuition that something has changed. This essay is what they tell us, afterwards, made the intuition legible. That is most of what a framework is for.