Scenario: The Industrialization of Knowledge Work (2026.1)
N.B. Each year, I create scenarios1 for the upcoming year—a tradition that began during my time as Tech Director at Elisa. This year, I will present these scenarios as detailed explorations before summarizing them. Here is the first2 and perhaps most crucial shift to watch for.
It begins with an admission: I was wrong.
Since 2017, we have adopted the Transformer model3, which is the key innovation behind the current surge in Generative AI. Typically, it takes 5 to 10 years for a technological advancement to have a significant impact on business or society. This delay occurs because humans adapt much more slowly than technology does.
By 2026, we will likely4 see the implications ripple through society, prompting me to correct my narrative.
I have been openly skeptical about “Agents” and the idea of a “Digital Workforce.” I questioned whether combining large language models, real-time search, and robotic process automation could create digital workers capable of replacing any human labor5.
While much of the marketing surrounding the digital workforce may be exaggerated, I now recognize that Agents do represent a significant structural advancement. They are not robotic workers, nor will they ever be. Nevertheless, the technology behind these “agents” will be fundamental to the industrialization of all knowledge work.
The Second Industrial Revolution?
To grasp the significance of this shift, we need to examine manufacturing. Before the 20th century, we built things using tools and relied on master artisans. The automotive industry didn’t truly explode until Henry Ford6 introduced the conveyor belt. Ford did more than enhance worker efficiency; he revolutionized the entire manufacturing process by industrializing it.
In the world of knowledge work, we have not yet experienced our Henry Ford moment. Although we have advanced from typewriters to word processors and from Google Docs to ChatGPT, these advancements are just tools—they are “optimizations.” They help individual workers perform their tasks more quickly, much as giving a blacksmith a powered hammer would.
Agentic architecture is the conveyor belt for knowledge work.
When I say “industrialize,” I don’t just mean speed. I mean a fundamental transformation in how we work. We are shifting from a model where humans use AI as a tool to one where the Agents serve as the core architecture, with humans managing the flow of operations.
An Agentic Operating Model?
My prediction for 2026—and the systemic shift we will see rise by 2027—is that the first companies will transition to Agentic Operating Models7, with a bang.
In these companies, the management layer itself becomes agentic. I’m not talking about automated decisions or shopping carts; I am referring to the orchestration of complex knowledge work—strategy, logistics, coding, or marketing operations. These processes will be run by an agentic—or perhaps, cybernetic—layer that vastly enhances the impact of both digital and human labor.
Why am I so confident this will happen? Capitalism.
In public examples of this kind of “industrialization”, we observe efficiency gains of not just 10% or 20%, but rather improvements in value creation ranging from 6 to 8 times.
No owner of a capitalistic enterprise can afford to overlook such a financial opportunity. It is simply unthinkable to ignore a potential seven-fold increase in impact8 and respond with, “No thanks, we prefer to stick to the old ways.”
What to Look For
This shift will inevitably force company boards to act. Investors will demand to know why one company operates with 1990s workflows while a competitor has embraced a progressive “Agent Mode”.
The transformation won’t happen overnight; the weight of legacy systems is significant. But if this scenario unfolds as I expect, the first signs will become apparent next year.
Don’t just focus on AI-native startups. I suggest you pay attention to bold legacy companies announcing shifts in their operating models9. Watch for those that move beyond merely discussing “copilots” for their employees to actively embracing the “industrialization” of their workflows.
This distinction is crucial: it is the difference between giving people faster hands and building the conveyor belt.
Footnotes
-
There are many more talented futurists than I. Still, I’d be happy to assist when exploring technology-related scenarios, especially with input from mainline strategy professionals and futurists. ↩
-
Initially, this was the second or third scenario I would address. However, after reevaluating the potential impact and the plausibility of its realization, I decided to move it up the queue. Just in case I end up dropping the ball on the other scenarios. ↩
-
While it might be more SEO friendly to refer to “LLMs,” or even just “AI,”, referencing the Transformer model is a personal preference. It highlights the specific technological breakthrough that has enabled the current wave of innovation. ↩
-
While these tools are now accessible and we can empirically validate their effects on a small scale, predicting the pace of change remains difficult. Specific issues may only emerge when implementing these tools at scale, and we may encounter challenges to adoption due to growing skepticism resulting from the current wave of overhype surrounding AI and Agents. ↩
-
In a study titled ‘How Do AI Agents Do Human Work? Comparing AI and Human Workflows Across Diverse Occupations’, Zora Zhiruo Wang et al. have adressed the issues in depth. While I disagree with some of the recommendations, the study does good work in highlighting the misalignment between marketing and reality. (https://arxiv.org/abs/2510.22780) ↩
-
I acknowledge that Henry Ford’s legacy is complex and controversial. This content focuses solely on Ford as an industrialist and a pioneer of mass production. ↩
-
Some people call this an “Agentic Operating System,” or “AI Native” I prefer to keep business discussions clear and avoid confusing them with artificial computer jargon. ↩
-
While stating effects on revenue or profits can create a stronger message and may be backed by public examples, citing such numbers is unwise, as case-level or business unit-level figures can not be reliably compared between two organizations. ↩
-
Additionally, I will be on the lookout for incumbents that suddenly increase personnel and training costs, indicating a move to upskill the workforce for the agentic age. ↩