A Jobless Boom – Anatomy of the jobless economic boom

By Francisco (Tony) Navarro-Sertich – Driving AI & Agentic Platform Shifts

Artificial intelligence will undoubtedly have a significant impact on the structure of the economy and the future of work. Much of the public conversation has focused on the risks of displacement, often framing the issue in binary terms: either AI augments labor or it replaces it. I suspect the reality will be more paradoxical. We may be entering a period of extraordinary economic growth alongside meaningful reductions in labor demand across large portions of the knowledge economy. In other words, a jobless boom.

This is not because demand will collapse, nor because firms will suddenly become less ambitious. Quite the opposite. The boom may occur precisely because organizations are about to gain access to a new class of productive capability, one that can increasingly perform cognitive work once reserved for humans. The question is not whether this will happen in isolated cases. It is how quickly it will happen, how broadly it will diffuse, and what new economic structures will emerge around it.

There are several forces converging to make this possible.

The first is that AI models have become materially more capable along a few defining dimensions. One is their ability to operate across longer time horizons. Increasingly, these systems are not limited to a single exchange or discrete prompt, but can sustain multi-step reasoning and execution across extended workflows. This enables them to take on longer-horizon tasks, in some cases compressing many hours of human work into a single agentic process, even if the results remain imperfect. Another is computer use itself. Systems can now manipulate what sits behind the screen with increasing fluency, navigating software environments, retrieving information, executing sequences and operating tools in ways that are beginning to rival, and in certain narrow contexts exceed, human speed and consistency. A third is steerability. Models are becoming more compliant, more tool-oriented and less prone to chaotic behavior than earlier generations. Rather than simply generating answers, they are increasingly capable of following structured procedures, using tools in an orderly way and staying within defined bounds.

The second force is the rise of context engineering and harness engineering. If models are the engine, these systems are the roads, controls and operational environment. They are what tie agentic operations together inside the enterprise. This matters because jobs are not monoliths. They are bundles of workflows, and workflows are bundles of tasks. Once this decomposition becomes explicit, organizations can begin to reassign portions of work to machines. We are slowly transitioning from a world in which technology is built primarily for humans to one in which technology is increasingly underpinned by machine operators acting on behalf of humans.

At first, firms will deploy AI into narrow domains. Over time, the more important shift will be contextual saturation. Models will increasingly be given access to the organization’s operating environment: its information repositories, data systems, permissions, memory, workflow tools and internal logic. If we think of an organization simply as a collection of people and systems coordinated around a common goal, then a model that can navigate across that pipeline of information becomes an extraordinarily powerful participant. Its greatest strength is not merely retrieval, but synthesis: the ability to connect information across silos, persist across time and transform scattered context into usable output.

A third force is the rise of the AI skill. Skills are important because they begin to package know-how in a portable and repeatable way. In essence, they specify how a model should behave within a given environment: what sequence to follow, which tools to use, what rules to obey and what a good output looks like. This is especially significant for tacit knowledge and knowledge work, which historically has been hard to systematize. Once such logic is captured in interoperable and deployable formats, organizations can replicate valuable work patterns quickly and with far less dependence on any one individual. Institutional knowledge begins to move from the heads of employees into the operating layer of the firm.

The fourth force is the emergence of the AI harness itself. A harness is more than an interface. It is the operating system that sustains, governs and provisions agentic workflows. It contains the rules, the security layers, the controls, the memory and the orchestration logic by which these systems function. In a sense, it plays a role analogous to that of management infrastructure for humans. It provisions capability, enforces standards, retains institutional memory and ensures that work is carried out within the boundaries of the organization. Over time, this layer may become one of the most important assets inside the enterprise, because it is where repeatability, compliance and scale converge.

This is how the jobless boom begins.

Today, a substantial share of organizational spend is tied directly to human labor. As more of that spend is captured through repeatable agentic processes, firms will see significant productivity gains. The same organization will be able to produce more output with fewer traditional labor inputs. Revenue may rise. Margins may expand. The pace of execution may accelerate. Yet the composition of inputs will change. Where firms once relied primarily on human knowledge workers, they will increasingly rely on providers across the AI stack: hardware, compute, infrastructure, models, tooling, applications and harnesses.

This is what makes the coming period economically powerful and socially disorienting at the same time. Growth does not disappear. It may even accelerate. But the labor intensity of that growth declines.

That said, this is not the entire picture. There is a silver lining, and it is a significant one.

Automation will not arrive everywhere at the same speed. In parallel with the compression of many white-collar functions, we are likely to see enormous investment in sectors that remain capital-intensive, physically grounded and difficult to automate end-to-end. Energy, semiconductors, data centers, industrial infrastructure, defense, logistics, agriculture and advanced manufacturing all stand to benefit from the buildout required to support the AI era. The AI revolution is targeting the knowledge worker first. But the infrastructure that sustains that revolution lives in the physical world.

This creates an important counterbalance. While many legacy white-collar functions in law, banking, financial services and administrative knowledge work may begin to shrink or transform, there may simultaneously be a broad resurgence in demand for engineering talent, technicians, specialists, builders and operators working across the world of atoms. In some respects, this may resemble a new industrial cycle, one that redirects economic energy from bits into the physical systems required to sustain those bits at scale.

This is why the notion of a “jobless economy” requires precision. It may be jobless in some sectors and booming in others. It may appear labor-light from the perspective of legacy office work while generating entirely new classes of demand elsewhere. The transition will not be smooth, and it will not be evenly distributed. Large portions of the population may require retraining, reskilling and support. Policy conversations should not lag this reality. They should be happening now, and they should be grounded not just in fear of displacement, but in a serious attempt to shape where opportunity will emerge next.

There will almost certainly be disruption. That much seems unavoidable. But disruption and optimism are not opposites. Schumpeter’s notion of creative destruction remains useful here: creation and destruction are often part of the same act. The decline of one labor structure may coincide with the rise of another. The challenge for society is not merely to predict this shift, but to govern it well enough that people can move through it with dignity and purpose.

I remain optimistic.

Not because the transition will be painless. It likely will not be. But because the underlying productive capacity being unlocked is extraordinary, and because human societies have repeatedly shown an ability to reorganize around new technological foundations. The task ahead is not simply to prepare for automation. It is to participate in shaping the economic, institutional and cultural systems that will sit on the other side of it.

A jobless boom may sound like a contradiction. It may soon become one of the defining economic realities of our time.