Realized

In plain language: What’s actually being put to work.

Definition

Realized is the companion perspective to Capacity. Where Capacity names the outermost Boundary — the maximum reach of the Agent’s sensing, modeling, and affecting — Realized names what the Agent is in fact putting to work. Realized can touch Capacity but can never exceed it. If Capacity is the solid outer boundary, Realized is the smaller shape inside it.

The gap between Capacity and Realized is the framework’s primary diagnostic space. When Realized falls short of Capacity, the Agent has capability it is not using. Whether that matters depends on the Agent’s objectives — per the Capacity is not Value principle, not every gap is worth closing. But naming the gap is the first step in determining whether it matters.

Realized is not tied to a single moment. The time window over which Realized is evaluated depends on the question the analyst is asking. “What did this team produce this quarter?” and “what has this VP accomplished in his tenure?” are both Realized questions — they differ in time slice, not in concept. A narrower slice reveals different patterns than a broader one: a VP’s Realized over this morning might be only email and meetings, while his Realized over this year includes regulatory analysis, financial modeling, and negotiation.

When the time slice narrows to a single moment, Realized becomes Actual Input (on the Computational Boundary) or Actual Output (on the Causal Boundary). Actual Input is the instantaneous signal crossing the Agent’s sensing channels right now. Actual Output is the instantaneous effect flowing through the Agent’s Causal Boundary right now. Both are specific forms of Realized — the general term applied at the narrowest possible time slice. The Cognitive Boundary has no named instantaneous form; what the Agent is modeling at a given moment is simply Cognitive Realization at that instant.

What Realized is not:

• Not Capacity. Realized is always less than or equal to Capacity, never greater. An Agent cannot put to work more than it has. When Realized appears to exceed Capacity, the diagnosis is that Capacity was measured wrong — the Agent had more than was initially visible.
• Not fixed. Realized changes as the Agent’s circumstances, attention, and choices change — typically faster than Capacity, because Realized can shift with a change in Frame, a new input, or a decision.

Relations

Realized is the companion perspective to Capacity. The gap between them is one form of Trapped Intelligence — capability that exists but is not being used.

Four categories of factors shape what an Agent’s Realized looks like at any given time:

• Frame: the Agent’s self-model determines which capabilities it recognizes and puts to work. When Frame is narrower than the Boundary, Capacity goes unused because the Agent doesn’t know it has it. This is the most developed cause in the framework and the one with the highest-leverage intervention — change Frame alone and Realized shifts without any change to Capacity or the world.
• Position in Topology: the Agent’s location in a Composition determines what signals reach it and what effects it can produce. Same Capacity, different position, different Realized.
• At Composition scale: cognitive overlap and interface integrity. Can the Agents in a Composition model each other well enough to coordinate? Can the coupling between their Boundaries carry the signals? When either fails, the Composition’s Realized falls below its Capacity.
• The Agent’s own choices. Within what Frame permits and Topology enables, the Agent selects which capabilities to exercise. A VP who knows he has regulatory expertise but focuses on negotiation is making a choice within his Realized, not suffering from a gap he cannot see.

Actual Input and Actual Output are the instantaneous forms of Realized on their respective Boundaries. Actual Input is Computational Realization at a single moment; Actual Output is Causal Realization at a single moment. Realized is the general term that subsumes both when the time slice narrows to a point.

Example — CEO

A CEO looks at his eight-person marketing department and sees what it produces: paid advertising campaigns, social media content, and event logistics. That is the department’s Realized — what it is actually putting to work.

He notices what is absent. Two of his marketers have strong data-analytics backgrounds, but neither is doing analytics work. One doesn’t know the company has a customer data platform — his Frame doesn’t include the Tool, so his Realized on that dimension is zero even though the Capacity is there. The other knows the platform exists but has never been asked to use it — his manager assigns him to event logistics because that is what the role description says. His Frame includes the tool; the organizational structure channels his effort elsewhere.

Same gap — analytics Capacity that is not being Realized — two different causes. The first is a Frame problem: make the marketer aware of the tool and his Realized expands immediately. The second is a position and choice problem: the role description and the manager’s priorities shape what gets put to work, and neither the marketer nor the manager has decided to redirect.

The CEO who sees only the gap (“we need to hire an analyst”) misses the diagnostic. The CEO who sees Realized as distinct from Capacity asks: “do we lack the capability, or do we have it and something is preventing it from being put to work?” The answer determines whether the intervention is hiring, awareness, or restructuring.

Example — Research

An AI agent — a language model with a provisioned set of Tools — is deployed as a customer-service assistant. Its Capacity includes: generating text, reasoning over complex queries, accessing a knowledge base of product documentation, looking up order status through an API, and issuing refunds through a payment system. Those capabilities are provisioned and available.

Its Realized is narrower. The agent answers product questions using the knowledge base and looks up order status reliably. But it has never issued a refund — the refund tool is provisioned in the deployment but not mentioned in the agent’s system instructions. The agent’s Frame doesn’t include the tool, so its Realized on the refund dimension is zero. The Capacity is there; the Frame gap keeps it unrealized.

A different gap appears at the knowledge base. The agent knows the knowledge base exists and queries it regularly, but its queries are shallow — it searches by keyword rather than using the semantic search feature. The agent’s Frame includes the knowledge base but not the full extent of its capability. Realized falls short of Capacity not because the tool is invisible, but because the Agent’s model of the tool is incomplete.

A configuration change — adding the refund tool to the system instructions and providing examples of semantic search — moves both capabilities inside the Agent’s Frame. Realized expands immediately. No change to the underlying Capacity, no change to the deployment, no new tools provisioned. Only the Agent’s self-model changed.

The diagnostic parallel to the CEO case: the deployment engineer asking “do we need to build a new tool?” is asking a Capacity question. The deployment engineer asking “why isn’t the agent using the tools it already has?” is asking a Realized question. The Capacity/Realized distinction tells him which problem he is actually facing.