# AI economics control plane kit

Use this kit after an AI workflow reaches real users. The purpose is to make investment decisions from operating evidence instead of launch assumptions.

## Control-plane scope

- Baseline: current volume, handling time, rework, delay cost, quality loss, escalation load, and labor mix.
- Adoption: eligible users, active users, assisted tasks, abandoned tasks, enablement events, satisfaction, and reviewer interventions.
- Unit economics: model spend, retrieval, tool calls, orchestration, monitoring, human review, exception recovery, support overhead, and avoided manual cost.
- Quality and risk: acceptance rate, reopen rate, policy exceptions, customer impact, reviewer disagreement, and control cost.
- Scale decisions: expand, tune, hold, or retire by workflow class and cohort.

## Operating cadence

Run monthly value reviews with finance, operations, product, risk, and workflow owners. Every review should end with a decision, owner, validation signal, and date.

## Decision rule

Do not expand an AI workflow when adoption, unit economics, quality, and control cost are not visible by cohort.