The Node Mix Over Time
Capacity split between human nodes and AI agent nodes, from the earliest data we have through 20 years forward. The solid region is history. The curved forward region is modeled from this organization's velocity, readiness, data posture, and regulatory constraints.
Human Time vs. System Time
The core arithmetic. A human node works a bounded year. An agent node runs continuously. This is a measure of capacity available to the mission, not a headcount-reduction plan.
(40 hrs x ~47 weeks)
(24 x 7 x 52)
-
before parallelism and before agents spinning up agents
Per-role economics
| Role | Human hrs/yr | Avg salary | Est. agent run cost/yr | Provenance |
|---|
Scenarios: Bear, Base, Bull
How much capacity gets added, and what it does to output. Three cases, every assumption visible. Forward figures are modeled.
The Multi-Layered Latency Stack
Eight layers of delay that an AI-Driven Enterprise compresses. Grey is where this organization sits today. Blue is where each layer realistically lands as agents are deployed. Governance keeps a floor: the goal is to reserve human review for high-stakes exceptions, not to remove oversight.
Latency scored 0 (near-instant) to 10 (crippling drag). Lower is better; the blue bars should sit left of the grey.