Browse Skills

ISO 42001 AI Lifecycle Procedures

    Get Started with This Skill

    Overview

    Draft AI lifecycle management procedures for ISO/IEC 42001:2023 (concept to retirement).

    Getting Started

    Add your source materials to the project workspace, then activate the ISO 42001 AI Lifecycle Procedures skill. The agent will guide you through the process step by step.

    Example conversation

    You: We need documented procedures for the full AI lifecycle—from concept and data through development, deployment, monitoring, incidents, and retirement.

    Agent: I’ll draft lifecycle procedures using the skill reference for each stage: concept & design; data acquisition and governance; model development; validation & testing; deployment; monitoring; incident response; retirement/decommissioning. For each I’ll define purpose, roles, inputs/outputs, and links to risk, data governance, and incident procedures. Then I’ll run lifecycle_procedures_check to confirm all stages are covered and fix any gaps.

    Output excerpt

    Lifecycle procedures excerpt

    Concept & design — Purpose: define requirements and governance for new or changed AI systems. Activities: requirements doc; impact assessment; design decisions and approvals. Outputs: approved concept; link to risk register. Owner: Product/AIMS.

    Data acquisition and governance — Purpose: ensure data used for AI meets quality, provenance, and lawful-use requirements. Activities: source assessment; quality checks; bias/gap assessment where relevant. Outputs: approved dataset; lineage. Owner: Data/Product. Ref: Data governance procedure.

    Deployment — Purpose: release to production with controlled rollout and roll-back. Activities: readiness checklist; approval; release; post-deploy checks. Outputs: deployment record; go-live sign-off. Owner: Product/Engineering.

    Monitoring — Purpose: ongoing performance, drift, and oversight. Activities: [e.g. weekly metrics, drift alerts, human oversight log]. Outputs: monitoring records; escalation to incident if needed. Ref: Monitoring procedure.

    Extension and validation

    The skill includes lifecycle_procedures_check, which validates that the procedure set covers all required lifecycle stages: concept & design; data acquisition and governance; model development; validation & testing; deployment; monitoring; incident response; retirement/decommissioning. Run it after drafting and address any missing stages.

    Ready to let your expertise drive the workflow?

    Stop wrestling with rigid templates and complex tooling. Write your process in markdown, let the agent handle the rest.

    Get Started