3 The economics of autonomy

Traditional return on investment (ROI) models were designed for a world of deterministic automation: systems whose behavior is fully specified at implementation and whose benefits accrue in a straight line. In that world, economic evaluation is relatively simple: automate a known volume of work, calculate labor savings, and project a payback period. Agentic systems break these assumptions. They operate under uncertainty, require continuous governance, and exhibit nonlinear paths to value. When applied to autonomous systems, traditional ROI models consistently produce business cases that look compelling on paper but fail in practice.

3.1 Case study: The claims-triage pilot that proved nothing

3.2 The Agentic Economics Canvas (AEC)

3.3 Lens 1: Cost structure: what autonomy actually costs

3.3.1 Build costs

3.3.2 Operating costs

3.3.3 Governance costs

3.3.4 Failure costs: the cost of autonomous error

3.4 Lens 2: Benefit structure

3.4.1 Coordination Cost Reduction (CCR)

3.4.2 Productivity and quality gains

3.4.3 Decision value

3.4.4 Option value

3.4.5 Realizing the ramp-up in benefits

3.4.6 Benefit prioritization by context

3.5 Lens 3: Risk adjustment (pricing uncertainty and failure)

3.5.1 How risk reshapes the economic profile

3.5.2 Operationalizing uncertainty: the three risk mechanisms

3.5.3 Risk-adjusted decision-making

3.6 Lens 4: Net Agentic Value (NAV) for determining economic viability

3.6.1 The NAV curve: from learning to scale

3.6.2 Interpreting NAV patterns

3.6.3 NAV as an ongoing governance tool

3.7 Applying the canvas: from analysis to decision

3.7.1 Comparative analysis: agentic vs. traditional automation

3.7.2 Sensitivity analysis

3.7.3 When each approach makes sense

3.7.4 Case study: Using AEC for a go/no-go decision

3.8 Anti-pattern: The pilot paradox

3.9 Pre-architecture gate: NAV validation toolkit

3.9.1 The architect's NAV validation checklist