The Economics of Delegation

Delegation has always been the fundamental economic act of scaling human output. When cognitive labour becomes infinitely delegable to AI agents, the economics of everything changes.

What happens when the bottleneck isn't "can I afford to hire someone" but "can I specify what I want clearly enough"? The cost of delegation drops to near-zero. The constraint shifts entirely to articulation and judgment.

The Historical Arc of Delegation

• Pre-industrial: delegation = slavery, servitude, apprenticeship. High coordination costs, limited by physical presence.
• Industrial: delegation = employment. Org charts. Management hierarchies. Taylor's scientific management. The factory as delegation machine.
• Knowledge economy: delegation = specialization + outsourcing. You don't do your own taxes, legal, accounting. Transaction costs (Coase) determine firm boundaries.
• AI era: delegation = prompting. The cost of delegating a cognitive task approaches the cost of describing it.

Each era brings new forms of leverage and new constraints.

What Changes When Delegation Costs Collapse

The Coasean Boundary Dissolves

Coase's theory of the firm: companies exist because internal coordination is cheaper than market transactions. When AI agents can execute tasks at near-zero marginal cost, the calculation flips. Why employ a team when you can delegate to agents on demand?

The firm becomes a "mission specification engine" rather than a "labour coordination engine."

The Middle Layer Compresses

Middle management exists to translate strategy into execution and aggregate information upward. If agents can do both translation and aggregation, what remains?

What remains:

• Judgment calls that can't be specified
• Relationship management (trust, politics, culture)
• Novel problem framing

The middle doesn't disappear - it gets radically compressed.

Leverage Becomes Unlimited for Individuals

A single person can now orchestrate research teams, content production, customer service, and operations through agent delegation. The solo operator with agent leverage can compete with teams. The "one-person unicorn" becomes structurally possible.

But: the constraint shifts to the person's capacity for specification and judgment.

The New Scarcities

When delegation is cheap, what becomes valuable?

Articulation - The ability to specify what you want with enough precision that agents can execute. Prompt engineering is the trivial version. Real articulation is: goals, constraints, success criteria, edge cases, values. The people who can articulate well become the new bottleneck. "I'll know it when I see it" becomes expensive - iteration costs attention.

Judgment - Agents can execute; humans must decide what's worth executing. Prioritisation, taste, ethics, strategy - these remain human. The market will increasingly pay for judgment, not execution.

Attention - Even with infinite delegation, human attention is finite. You can only review so many outputs, make so many decisions, maintain so many agent relationships. Attention becomes the ultimate scarce resource. Previously, attention was scarce because content was abundant. Now, attention is scarce because work is abundant.

The Delegation Paradox

More delegation capacity → more things get attempted → more review required → attention exhausted → quality drops.

The paradox: infinite leverage without infinite attention creates its own ceiling.

Solutions:

• Hierarchical delegation (agents reviewing agents) - requires trust infrastructure
• Taste compression (specifying preferences so tightly that review becomes sampling)
• Outcome orientation (measure results, not process)

Economic Implications

Labor Markets

Tasks that can be specified cleanly: price collapses to agent marginal cost. Tasks that require judgment, relationship, or novelty: price premium grows.

The "premium on human" widens. But most current jobs are mixtures. The job doesn't disappear; it fragments. The specification parts get delegated; the judgment parts get concentrated.

Wage polarisation accelerates. The people who orchestrate earn more. The people who did the delegatable parts... transition or exit.

Pricing Models

Shift from time-based (hours, salaries) to outcome-based (deliverables, results). When execution is cheap, charging for execution time makes no sense. The value is in the specification and judgment, which is hard to price directly.

Rise of "outcome as a service" - you pay for leads, not hours; for code deployed, not developer time.

Capital Allocation

Human capital was valuable because training took years and execution capacity was limited. With delegation, the value shifts to the system that enables delegation.

Capital flows to:

• Agent infrastructure (the "plumbing")
• Data moats (agents need information)
• Trust/reputation systems (delegation requires confidence)
• Specification tools (helping humans articulate)

Inequality

Early delegation advantage compounds. Those who can orchestrate agents early capture disproportionate value. This is familiar from every technology wave, but the speed is different.

The question: will AI tools lower the barrier to orchestration fast enough to distribute the gains?

The Organisational Response

Small becomes viable - 2-person companies with agent armies can deliver what required 50 before. This changes competitive dynamics: incumbents face more agile competitors.

Management becomes curation - The manager's job shifts from directing humans to curating agent outputs. Performance reviews → output audits. Team meetings → system reviews. Hiring → agent selection and configuration.

Culture becomes specification - Company culture is the unwritten specification of "how we do things." With agents, culture must become explicit enough to encode. "We value quality over speed" → specific rubrics, gates, evaluation criteria. The companies that can articulate their culture win.

What This Means for Individuals

Career advice in the delegation era:

1. Develop articulation skills - can you specify what you want clearly?
2. Develop judgment - can you evaluate outputs and make good decisions?
3. Understand the tools - not coding, but orchestration
4. Focus on the non-delegatable - relationships, trust, novel problems
5. Build attention discipline - your attention is your scarcest resource

The new status markers:

Previously: team size, budget, title. Now: leverage ratio (output / personal attention).

The flex is "I did this with three agents and two hours of my time."

Open Questions

• Does infinite delegation lead to infinite entrepreneurship, or does it consolidate power among the best orchestrators?
• How do we price judgment when it's the bottleneck but hard to measure?
• What happens to the social fabric when "having a job" becomes optional for production but necessary for identity?
• Will people delegate things they shouldn't? (Relationships? Creative expression? Judgment itself?)
• Does the ability to delegate make us better at articulation, or atrophy it?

The economics of delegation is the economics of leverage. Every technology that made delegation cheaper - writing, printing, computing, networking - reshaped economic structure. AI is the first technology that makes cognitive delegation cheap.

The question isn't whether this changes everything. It does. The question is: who captures the value, and how fast does it redistribute?