Agentic Labour: When Work Becomes Executable

Software reduced coordination costs.

Platforms reorganized production.

Work became modular.

The next shift is subtler:

Some forms of labor are becoming executable.

Not automated in the narrow, repetitive sense — but structured as goal-oriented workflows that can be delegated to autonomous systems.

This introduces a new layer into institutional design.

From Tasks to Execution Loops

Traditional automation executes predefined steps.

Agentic systems operate differently.

They:

• Accept goals
• Select tools
• Evaluate intermediate results
• Iterate until completion

Work becomes a loop.

This structure resembles how human operators perform knowledge work.

The difference is that execution can occur without continuous human intervention.

Labor as Workflow

Historically, labor was tied to roles.

With digital coordination, it became tied to tasks.

With agentic systems, it becomes tied to workflows.

A workflow:

• Has defined inputs
• Produces measurable outputs
• Can be monitored
• Can be replayed
• Can be improved

Once a workflow is formalized, it can potentially be executed by:

• A human
• A human assisted by software
• An autonomous agent

The unit of economic value shifts toward execution capability.

Coordination Costs Shift Again

If agents reduce the cognitive overhead of:

• Research
• Drafting
• Monitoring
• Allocation
• Reporting

Then certain internal hierarchies may become thinner.

At the same time, systems capable of managing large numbers of agents may gain coordination advantages.

The cost comparison between:

• Internal execution
• External human contractors
• Agentic systems

becomes more complex.

But Agentic Execution Introduces New Costs

As with each previous coordination shift, the reduction in some costs is accompanied by the emergence of others.

• Verification and oversight costs — agentic outputs require evaluation. The faster and more autonomous the execution, the more expensive it becomes to confirm that outputs are correct, safe, and aligned with intent.
• Failure-mode costs — agents operating in loops can compound errors at machine speed. Recovery from cascading mistakes may be more expensive than the productivity gains from autonomous execution.
• Liability and attribution costs — when an agent executes a workflow that produces a regulatory breach or a contractual violation, determining responsibility is structurally harder than in human-executed processes.
• Infrastructure lock-in costs — firms that build workflows around specific agentic platforms create dependency on those providers. Switching costs may reintroduce the kind of asset specificity that Williamson identified as a driver of vertical integration.

The boundary of the firm is affected once more — but not in a single direction.

Hybrid Labor Structures

Agentic labor does not eliminate human labor.

It changes its composition.

Humans may focus more on:

• Strategy
• Oversight
• Exception handling
• Relationship management
• Institutional judgment

Agents may handle:

• Structured analysis
• Repetitive coordination
• Information synthesis
• Procedural execution

This creates hybrid teams.

Not firms without people. But firms with different ratios of human to executable work.

Measurement and Traceability

Agentic workflows produce traces.

Execution steps can be logged, audited, and replayed.

This has implications for:

• Accountability
• Quality control
• Performance measurement

When work becomes executable, performance becomes inspectable at finer granularity.

Measurement systems adapt accordingly.

As discussed earlier in the series, metrics shape behavior.

Agentic labor introduces new measurable surfaces.

Agentic Labor at the Edge

For independent professionals, agentic systems may function as capability multipliers.

A single operator can:

• Run parallel workflows
• Maintain higher throughput
• Serve more clients

This amplifies fractional work models.

At the same time, firms that centralize agentic infrastructure may scale coordination more rapidly.

Institutional outcomes depend on:

• Tool accessibility
• Capital concentration
• Governance structures
• Regulatory adaptation

The direction is not predetermined.

A Structural Variable, Not a Replacement Narrative

Agentic labor is best understood as a coordination variable.

It reduces certain costs — cognitive overhead, execution latency, the marginal cost of structured work.

It increases others — verification burden, failure-mode risk, liability complexity, and infrastructure dependency.

It shifts capability distribution — amplifying individuals with access to agentic tools, while potentially reinforcing scale advantages for firms that can absorb the new overhead costs.

Institutions evolve in response to these shifts.

Just as the managerial corporation emerged under high coordination cost, and platforms emerged under lower digital coordination cost, agentic labor represents another adjustment in how execution is organized.

The net institutional effect depends on which cost movements dominate. In domains where verification is cheap and workflows are well-defined, agentic execution may favour smaller, leaner structures. In domains where liability, compliance, and failure costs are high, it may reinforce the need for organisational scale.

The firm remains.

The boundary moves.

Work adapts.

In the next post, we examine a dimension that coordination cost alone does not capture: how knowledge moves between actors — and why firms have historically served as containers for that diffusion.