Velocity Recalibration Framework

A structured approach to recalibrating sprint velocity when a team adopts AI tools. Prevents the over-commitment and stakeholder misalignment that results from treating AI-inflated velocity as sustainable capacity.

Syntax

scrum-master-ai

Velocity Recalibration Protocol

Phase 1: Establish AI Contribution Baseline (Sprints 1-3 of AI adoption)
- Tag stories as: AI-heavy (>50% AI contribution), mixed (20-50%), or minimal (<20%)
- Track velocity separately for each category
- Do NOT use pre-AI velocity as a planning input

Phase 2: Separate Capacity Model (Sprints 4-6)
- AI-heavy story capacity: [observed velocity for AI-heavy stories]
- Mixed story capacity: [observed velocity for mixed stories]
- Human-primary story capacity: [observed velocity for minimal AI stories]
- Note: human-primary stories exist because some work requires domain knowledge AI lacks

Phase 3: Calibrated Planning (Sprint 7+)
- Plan sprints using the category-specific velocity, not aggregate velocity
- Add leading indicators: cycle time, rework rate, escaped defects
- Review and update the model every 4-6 sprints

Example

scrum-master-ai

// Team recalibration after 6 sprints:

preAiVelocity: 42, // story points per sprint

aiAdoptionVelocity: {
  sprint1: 61, // spike from AI novelty
  sprint2: 58,
  sprint3: 71, // high AI-heavy stories
  sprint4: 54, // harder stories, AI less helpful
  sprint5: 63,
  sprint6: 59
}

calibratedCapacity: {
  aiHeavyStories: "22 points per sprint (well-specified, bounded stories)",
  mixedStories: "24 points per sprint",
  humanPrimaryStories: "16 points per sprint (architecture, discovery, novel integration)"
}

stakeholderNote: "Aggregate velocity increase is real but not uniformly sustainable. Plan by story type."