When we talk about AI’s impact on work, we often jump straight to a big question: which jobs will disappear? But a recent observation made me think the more important change might not be whether certain positions get eliminated entirely, but rather how the fundamental units of work are being transformed.

I recently spoke with two small business owners and startup founders. Both mentioned an interesting phenomenon: when hiring, they’ve started prioritizing questions not just about whether candidates can use AI tools, but about how large an AI agent team they can build, how large a team they can manage, and whether they have practical experience doing so.

The sample size is obviously too small to draw conclusions about the entire labor market. But it serves as a clear early signal: companies’ conception of ‘talent’ may already be shifting. When interview questions move from ‘what can you do’ toward ‘what can you orchestrate,’ careers themselves may be getting restructured.

Work Gets Decomposed Before Careers Get Recomposed

We typically understand work as complete positions: product manager, engineer, designer, analyst, operations, marketing. This understanding carries an assumption—that the primary responsibilities within each role remain bundled together, sufficient to constitute a relatively stable unit of work.

But AI may be changing this premise. Rather than entire positions disappearing first, AI might first decompose the work within those positions: research, synthesis, drafting, analysis, coordination, validation, response, tracking—these responsibilities that were once bundled together are beginning to be redistributed among people, tools, and agents.

Once the fundamental unit of work begins to change, it becomes difficult to maintain the traditional imagination of a career as ‘one job title corresponding to one stable role.’ What the market may care about more in the future is not just which position you belong to, but whether you can reassemble a set of capabilities into a productive work system.

If these founders’ observations are representative, what we’re seeing isn’t simply ‘AI skills are important,’ but something more profound: companies are beginning to view candidates as potential system designers. The question is no longer just whether you understand a particular tool, but whether you can design workflows, build agent teams, manage handoffs between humans and agents, and turn this into sustainable productive capacity.

This expectation is new because it layers together several previously distinct types of capabilities: you need to understand both implementation and management; both tools and structure; both how to complete tasks and how to decompose tasks, redistribute them, and integrate the results back together.

Careers May Become Recomposable Capability Modules

If the fundamental unit of work is truly changing, then careers will also struggle to maintain the traditional model of ‘one job title corresponding to one stable role.’ What the market may care about more in the future is not just which position you hold, but whether you can reassemble a set of capabilities into a productive work system.

This implies several possible changes:

• Careers may no longer be paths toward specific titles, but rather shifts from one capability combination to another that better fits current demands.
• Valuable people may not be those with the strongest single-point skills, but those best able to design and manage hybrid work systems—letting humans and agents each leverage their comparative advantages to produce greater value together.
• Work competitiveness may gradually shift from ‘how much can you do’ toward ‘how much total output can you amplify.’ The ability to systematize, scale, and repeat execution becomes another form of value. These three shifts are still very early-stage, but they at least give us a framework for re-examining our positions: rather than asking whether AI will take my job, we might ask how my work units are being redesigned, and what role I can play in this new structure.

This Change Is Still Early: Signal, Not Destiny

I’m not ready to write these observations as conclusions that the entire labor market has already shifted. The sample size is too small (n=2), and the timing is still too early. The tone of this article should be: change has begun, but the endpoint remains unclear. Not packaging early signals as grand prophecies, but honestly pointing out that the way work is divided may be changing.

If this line of thinking holds, then the next question worth asking isn’t ‘will AI take jobs away,’ but rather: when work is decomposed, how should people redesign their positions? What this article can do, perhaps, is simply to clearly pose this question and make it a starting point for subsequent thinking.

Extended Observation: Optimal Company Scale May Also Be Changing

An observation worth adding: if the fundamental units of work can be decomposed, then the optimal scale of ‘company’ as an organizational form may also shift.

We’re already beginning to see the emergence of ‘one-person companies.’ But one person often struggles to cover all the distinct aspects of a business—product R&D, technical implementation, sales, marketing. Although AI can assist at each stage, in a competitive market, ‘doing better than others’ still requires human judgment and specialization.

Thus, one possible future scenario: company scale converges to around 6-8 people, each specializing in different domains (product, design, technology, marketing), each using AI to amplify output while maintaining tight collaboration. Such teams could support substantial business scale without bearing the structural costs of large organizations.

From this perspective, large companies’ competitive advantages may be weakened—except in industries that must rely on large frontline workforces (like food service). In creative products or e-commerce, ‘small but focused’ teams with AI may be more flexible and cost-effective than hierarchically complex large organizations.

This remains, of course, an early signal. Not all industries will change at the same pace—public institutions and traditional industries may move much more slowly. But at least, behind the ‘careers being re-sliced’ phenomenon, there may be an implicit ‘organizational forms being redesigned’—these two threads are moving simultaneously.