Maturity Level 4 of 6

Integrated

Cross-functional integration and portfolio management.

Level 4 (Integrated) is where GenAI stops being "a program" and becomes "how the business runs" — embedded in every business unit's planning cycle.

Framework v2026.1 · Updated · machine-readable spec

What this level looks like in practice

GenAI is a portfolio, not a project list. Multiple business units run their own AI initiatives against a shared platform. Governance is standardized across the org.

Level 4 across all six dimensions

  • Strategy & Leadership

    GenAI embedded in every business unit's annual planning cycle. Portfolio management with stage-gates.

  • Data & Infrastructure

    Feature store or equivalent shared substrate used across teams. Data quality SLAs enforced.

  • Use Cases & Applications

    Formal portfolio with stage-gate reviews. Deprioritization is routine.

  • Talent & Culture

    AI literacy assessed as a competency. AI product management is a recognized track.

  • Governance & Risk

    AI system inventory mapped to EU AI Act risk tiers. Model cards standard.

  • Agentic AI

    Standardized tool-use protocol (MCP) adopted across multiple agents. Agent eval harness distinct from LLM evals.

Common blockers to the next level

  • Human-in-the-loop everywhere — autonomy is still a scary word.
  • No continuous evaluation — production quality only measured on release.
  • Reasoning models still treated as expensive novelty.

Worked example — Retail

A global specialty retailer has 12 GenAI systems in production across merchandising, marketing, and store operations. All use a shared vector store, shared guardrails, and shared observability. Each business unit owns its own use cases; the central AI platform team owns the substrate. Portfolio ROI is reviewed quarterly by the CEO.

For developers

Call this framework and its tools from your own agent via the Model Context Protocol (MCP) server. Works with Claude Desktop, Cursor, Zed, Continue, and the OpenAI Agents SDK.