Maturity Level 3 of 6

Operational

First production deployments with measurable outcomes.

Level 3 (Operational) is the hardest transition — from "we have pilots" to "we ship AI to production and measure the outcomes."

Framework v2026.1 · Updated · machine-readable spec

What this level looks like in practice

At least one production AI system with measurable business impact. Runtime guardrails in place. This is where the org starts to build durable capability rather than one-off pilots.

Level 3 across all six dimensions

  • Strategy & Leadership

    Published GenAI strategy tied to business outcomes. Cross-functional steering committee.

  • Data & Infrastructure

    Governed data pipelines feed production AI. Vector store is a shared internal service.

  • Use Cases & Applications

    At least one production deployment with measurable business outcome. Prioritized backlog.

  • Talent & Culture

    Role-based training curriculum. Dedicated AI/ML engineering function.

  • Governance & Risk

    Every production deployment passes risk/bias/compliance review. Runtime input/output filtering.

  • Agentic AI

    At least one agent in production with runtime guardrails and task-success metrics.

Common blockers to the next level

  • AI capability lives in one central team; other business units cannot self-serve.
  • Governance runs as a gate at launch, not a continuous discipline.
  • No portfolio-level ROI reporting — each use case is measured in isolation.

Worked example — Healthcare

A regional health system has deployed an AI-assisted clinical documentation tool across 3 specialties, measured a 22% reduction in after-hours charting time, and published the methodology internally. Runtime guardrails redact PHI before external API calls. Every deployment passes a documented clinical safety review.

Related on this site

Next steps