Skip to main content

Measuring Predictability

Organizations often measure:

  • Velocity
  • Story points
  • Cycle time

AI-driven organizations should additionally measure:

  • Architectural Consistency: Do generated features follow the same structure?
  • Reusability: Are generated components reusable?
  • Test Coverage: Are tests generated consistently?
  • Compliance: Does generated code follow skills?
  • Refactoring Frequency: How often must generated code be reworked?
  • Predictability: How consistent is the AI in generating features?
success

Predictability can be measured.

Mini Exercise​

Pick one metric you could use to judge whether AI-generated code is becoming more predictable.

Example​

A simple metric is structural consistency: how often generated files land in the expected folders with the expected names.