303: Who Gets to Be Autonomous?
The Imagination Crisis of the Agentic Era
The most powerful builders in history are alive right now, and most of them are still building within a very small imagination.
Not small in technical ambition. The systems are extraordinary. Agents that can reason, coordinate, generate, and act. Infrastructure that can scale across domains, absorb complexity, and operate continuously. For the first time, the act of building is no longer tightly constrained by capital, institutions, or even formal expertise. A single builder can now assemble systems that behave like organizations.
The constraint is elsewhere.
It sits in the frame through which builders decide what is worth building.
Social media reshaped that frame long before agents arrived. It built a maze of attention—an environment where visibility became value, and behaviour was continuously tuned through feedback. Builders learned inside that maze. They learned how to capture attention, how to retain it, how to convert it. Success became measurable, comparable, optimizable.
Agentic systems now sit on top of that architecture. If social media organized attention, agentic AI begins to organize intention. It can anticipate needs, simulate outcomes, guide decisions. It does not simply host activity; it begins to shape direction. In that sense, it functions less as a tool and more as an environment—one that influences how builders perceive the future itself.
This is where the imagination narrows.
When the surrounding system rewards extraction, optimization, and control, those become the default problems to solve. Much of what is being built today, despite its novelty, follows that path. Agents that accelerate workflows. Systems that capture more value. Platforms that scale coordination without redistributing power. The language remains familiar: productivity, efficiency, growth.
Yet something else has entered the vocabulary, and it carries a deeper contradiction.
Autonomy.
The agentic era is saturated with the term. Autonomous agents. Autonomous systems. The implication is that something fundamental has shifted—that we are moving toward systems that can act independently, perhaps even liberate us from constraint.
But autonomy, in any meaningful sense, has never been a defining feature of the system these builders operate within.
Workers have long operated within structures they did not design. Managers have been constrained by financial imperatives they do not control. Even founders navigate investor expectations, market pressures, and competitive dynamics that sharply delimit their choices. The system coordinates behaviour by structuring dependence, not by distributing autonomy.
So when builders describe agents as autonomous, what is being claimed?
At the technical level, autonomy means the ability to execute tasks without direct supervision. An agent can plan, decide, and act within a given scope. But the scope is defined elsewhere. The objectives are assigned. The evaluation criteria are external. What emerges is not autonomy in the fuller sense, but delegated execution.
This is the paradox at the center of the agentic imagination.
We are building systems described as autonomous within a culture that has never prioritized autonomy for the people inside it. The word migrates from a political condition to a product feature. It signals independence while remaining tethered to the structures that define its purpose.
The consequences are already visible.
An agent deployed within an extractive system will optimize extraction. An agent embedded in a hierarchical structure will reinforce hierarchy. Efficiency gains will be captured before they are shared. The system becomes more capable without becoming more accountable.
The language promises one thing. The structure delivers another.
This is where the role of the builder begins to shift, whether acknowledged or not.
To decide what to build is to decide how a system will behave, who it will serve, and how it will distribute capacity. These are governance decisions. Interfaces determine who can act. Data structures determine what can be seen. Incentives determine what is rewarded. In the agentic era, these decisions extend further. Builders are no longer just creating tools; they are constructing environments that coordinate both human and machine actors.
The line between software and institution continues to blur.
Yet the governance imagination has not kept pace with the technical one. Much of what is being built assumes a passive user, a centralized operator, and a model of value that accumulates upward. The agent may act independently, but the human remains constrained.
This is the inversion that defines the moment: machines framed as autonomous, people left within systems of dependency.
It does not have to resolve that way.
There are early signs—still marginal, but instructive—of systems designed with a different orientation. Open models that allow inspection and modification. Cooperative infrastructures that distribute decision-making. Agentic tools that scaffold collective deliberation rather than replace it. These efforts begin from a different premise: that the purpose of building is not simply to extend capability, but to expand participation.
Here, autonomy is treated less as a property of the system and more as a condition the system helps produce.
That shift changes the design problem.
It asks builders to consider not only what an agent can do, but what people can do because the agent exists. Can users understand how decisions are made? Can they influence those decisions? Can they refuse, redirect, or repurpose the system? Can communities govern the tools they depend on?
These questions are harder to optimize. They do not map cleanly onto existing metrics. They often move against prevailing incentives. Transparency can reduce control. Shared governance can slow execution. Distributed ownership can limit capture.
And yet, without these considerations, autonomy remains a surface effect—visible in the behaviour of agents, absent in the experience of those around them.
The deeper divide is not ideological. It is architectural.
Some systems concentrate decision-making and obscure their logic. Others distribute participation and expose their processes. Some treat users as inputs to be optimized. Others treat them as participants in an evolving system. These are choices made in design, long before they are debated in theory.
Open systems have begun to articulate an alternative logic. Not as a slogan, but as a practice of building. When code, models, and methods are shared, coordination becomes more collective. Authority becomes more contestable. The system remains imperfect, often unstable, but it expands the field of what can be imagined.
It also reframes the stakes.
If agentic technologies become the infrastructure through which decisions are made, then the design of those systems will shape how authority is exercised. Who defines goals. Who can intervene. Who understands what is happening and why. These are not abstract concerns. They are the operating conditions of everyday life.
What makes this moment distinct is that these conditions are being set, piece by piece, by builders.
Often implicitly. Often under pressure. Often guided by the same narrow signals that defined the previous era.
The horizon remains open, but not indefinitely.
As systems stabilize, they harden. Standards emerge. Habits form. The early assumptions—about autonomy, about value, about participation—become embedded. Changing them later becomes more difficult, more costly, more contested.
This is why the question of imagination matters now.
Not as an appeal to creativity, but as a recognition of constraint. Builders are working with unprecedented capacity inside inherited frames. The risk is not that they fail to build powerful systems. It is that they succeed in building systems that extend the logic of the present into the future.
The agentic era expands what can be built. It also expands the consequences of building.
Autonomy has entered the language. The systems will determine what it comes to mean.
Whether it remains a feature of machines, or becomes a condition experienced more broadly, will depend on how far builders are willing to step outside the frames that taught them what building is for.
The tools are already here.
The question is whether the imagination that guides them can keep up.
Agentic Design: A manifesto of care, agency, and connection.
Tiktok failed to load.Enable 3rd party cookies or use another browser