AI Governance Institute
All governance templates →How do we govern AI models from preview release through retirement?

Implementation Kit

AI Model Lifecycle Stage Gates and Retirement Checklist

Stage gates from preview to retirement. Stage definitions with promotion criteria, a register of events that force re-assessment, a promotion gate checklist, and a retirement checklist.

Who this is for: The model governance owner defining what has to be true for a model to advance, and what pulls it back for review.

Download the kit (Markdown) ↓4 artifacts. Every table also copies as CSV.

1. Lifecycle stages and promotion criteria

Spreadsheet

The stages your models move through and the evidence required to move between them.

Template

StageWhat it meansPromotion criteria to enter this stage
PreviewAvailable to a small internal group for evaluationRegistered; owner assigned; intended use documented
Approved for TestingWider internal use, non-production data pathsPreview eval complete; risk tier assigned; known limitations documented
Production (GA)Live, affecting real decisions or usersFull evaluation (incl. bias where people are affected); human oversight design; Legal sign-off for High tier; rollback plan
Under Re-assessmentStill live, but flagged for reviewA re-assessment trigger has fired
DeprecatedDiscouraged, migration underwayReplacement identified; migration plan with dates
RetiredOffRetirement checklist complete

Worked example

StageEnteredEvidence at promotion
Preview2026-01-10Registered; owner D. Portillo; intended use documented
Approved for Testing2026-01-24Preview eval done; risk tier High; known limitations logged
Production (GA)2026-02-11Full eval incl. adverse-impact; oversight design; Legal memo; rollback to v2.3.1
Under Re-assessment2026-08-30Base model changed VendorLM-3 to VendorLM-4
Production (GA)2026-08-30Eval delta reviewed (ratio 0.85 to 0.88); cleared same day

Acceptance criteria

  • The stages match how your organization actually ships models, not a generic template.
  • Each promotion criterion is objective evidence, not a judgment call.
  • Every production model has a current stage recorded in the registry.

2. Re-assessment trigger register

Spreadsheet

The events that move a live model to Under Re-assessment before it can continue.

Template

TriggerThresholdWho monitorsAction on fire
Model or base-model changeAny weights, prompt, or parameter changeMLOpsNew version; re-run eval before continuing
Performance drift<metric past its control limit>MonitoringRe-assess within X days
Bias metric shift<fairness metric past threshold>MonitoringImmediate re-assess; consider pause
Use-case or population changeAny change to what it decides or who it affectsBusiness ownerFull re-classification
New or changed regulationA regime now applies or its requirements changedLegalRe-assess against new requirements
IncidentAny Sev-1 or Sev-2 involving this modelIncident processRe-assess as part of post-incident review
TimeNo review in <N> monthsAI Gov LeadScheduled re-assessment

Worked example

TriggerThresholdWho monitorsAction on fire
Base-model changeVendorLM-3 to VendorLM-4MLOpsFired 2026-08-30; eval re-run, cleared
Bias metric shiftAdverse-impact ratio below 0.80MonitoringNot fired; current 0.88
Time90 days since last reviewAI Gov LeadNext scheduled 2026-11-30

Acceptance criteria

  • Every trigger names who monitors for it and what happens when it fires.
  • Thresholds are numeric where the trigger is a metric.
  • A fired trigger is recorded against the model with a date and an outcome.

3. Promotion gate checklist

Spreadsheet

Run before a model advances a stage. Same checklist, applied at each transition.

Template

CheckTestingProductionEvidence
Registered with all required fieldsYesYes
Risk tier assignedYesYes
Evaluation appropriate to the stagePreview-levelFull, dated
Bias evaluation (if people are affected)OptionalYes
Human oversight design documentedRecommendedYes for High
Rollback planRecommendedYes
Legal sign-offIf triggeredYes for High
Owner accepts stage accountabilityYesYes

Worked example

Promotion: Approved for Testing to Production, fraud-scoring model.

CheckStatusEvidence
Registered, risk tier assignedYesHigh
Full evaluation, datedYes2026-09-01
Bias evaluation (people affected)Yesno disparity above threshold
Human oversight design documentedYesanalyst reviews all declines
Rollback planYesrevert to rules engine
Legal sign-offYes
Owner accepts stage accountabilityYesHead of Risk Ops
DecisionPromoted2026-09-03

Acceptance criteria

  • The gate is run and recorded at every stage transition, not just the first.
  • Each check has attached evidence, not a tick.
  • A missing check blocks promotion absent a recorded exception.

4. Retirement checklist

Spreadsheet

Clean decommissioning: weights, logs, dependents, sign-off.

Template

StepOwnerDoneEvidence
No production traffic or scheduled jobs remainY / N
Dependent systems migrated or sunsetY / N
Weights deleted or archived per data policyY / N
Logs and evaluations archived to retention storeY / N
Registry status set to Retired with dateY / N
Inventory and documentation updatedY / N
Owner sign-offY / N

Worked example

StepOwnerDoneEvidence
No traffic or jobsMLOpsY30-day zero-call log
Dependents migratedMLOpsY2 pipelines moved to the new model
Weights archivedMLOpsYCold storage; prod deleted
Logs archivedAI Gov LeadY7-yr retention store
Registry status RetiredAI Gov LeadY2026-09-20
Inventory updatedAI Gov LeadY
Owner sign-offHead of Risk OpsY2026-09-20

Acceptance criteria

  • Every step has an owner and evidence before retirement is called complete.
  • Weight deletion or archival follows the organization's data retention policy.

Governance controls this kit produces evidence for

Completing the artifacts above gives you a head start on the evidence requirements for these controls.

MGV-001
MGV-001

The stage definitions and promotion criteria are the preview and staged-release policy.

CHM-002
CHM-002

The promotion gate checklist is the deployment gate process, applied at each transition.

CHM-005
CHM-005

The retirement checklist is the model deprecation and decommissioning procedure.

MGV-002
MGV-002

Stage entry criteria plus the gate checklist form the intake and approval workflow across the lifecycle.

MON-005
MON-005

The re-assessment trigger register defines the continuous-evaluation events that force a fresh review.

This kit backs one playbook. Read the full guidance for the reasoning behind each artifact.

Decide what to implement next

Assess your governance gaps, then create an action plan with owners and target dates. Build and export without an account; sign in when you want to save your plan.

Start the AI governance assessment →