Dependency map — AI Operating Model

The twelve dimensions of the maturity model are not a checklist — they form a directed graph. Strategic posture flows into AI-to-Value and decision rights. Agentic autonomy requires decision-rights maturity, which gates reversibility, which gates incident response, which surfaces observability evidence back to risk management. Data quality gates every decision-requiring dimension. Workforce capability gates the HITL meaningfulness that makes decision rights real. This map shows the load-bearing edges: which dimensions constrain which, and which gaps will cascade.

v1.0 Tier A simplification. The full 35-sub-dimension dependency spine lands in the Tier B PDF version (v1.0 and beyond). This diagram shows the dimension-level dependencies that hold across all three pillars.

AI Operating Model dependency map: twelve dimensions across three pillars (Strategy, Responsible & Agentic AI Governance, Enablement) with sub-dimension spokes and load-bearing cross-dimension handshakes.

Pillar 1 Strategy

D1 Strategy & Leadership

1.1Exec sponsorship
1.2Board oversight
1.3Strategy ↔ P&L
1.4AI FinOps
1.5Capital allocation
1.6Strategic posture ★

D2 AI-to-Value

2.1Use-case discovery
2.2Stop/pivot evidence
2.3Value realisation
2.4Build/buy/partner
2.5Portfolio sequencing

D3 Data & Knowledge

3.1Data lineage
3.2Data-product ownership
3.3PII & AU residency
3.4Training-data provenance
3.5Knowledge catalogue

D4 Tech & Platform

4.1Compute & serving
4.2Platform resilience
4.3Integration arch
4.4MLOps / LLMOps
4.5EvalOps

Pillar 2 Responsible & Agentic AI Governance

D5 Agent Architecture

5.1Agent identity (NHI)
5.2Lifecycle 5→500
5.3Eval methodology
5.4Multi-agent + HITL
5.5Runtime + BYOA
5.6Multi-vendor portability
5.7Autonomy + blast radius

D6 Decision Rights

6.1Decision-rights matrix ★
6.2Override & pause
6.3Accountability fwk
6.4Reversibility, redress ★
6.5Audit trail

D7 Ethical AI & Risk

7.1Ethics committee, AIA
7.2Bias, fairness, explain
7.3AU Ethics Principles
7.4AU reg mapping
7.5AI Impact Assessment

D8 Vendor Governance

8.1Vendor due diligence
8.2AI contract SoR
8.3Ongoing vendor perf
8.4Incident-notice SLAs
8.5Concentration risk

D9 AI Security

9.1DLP for agents
9.2Prompt-injection defence
9.3Security observability
9.4Shadow / BYOA blocking
9.5Agent IAM (NHI)

D10 Observability

10.1Incident classification
10.2Incident response
10.3Performance anomalies
10.4Data lineage runtime
10.5Notification triggers ★

Pillar 3 Enablement

D11 Workforce & HITL

11.1Tiered AI literacy
11.2HITL meaningfulness
11.3Reskilling pathways
11.4Champions & CoP
11.5Workforce-AI trust

D12 Adoption & Change

12.1Adoption metrics
12.2AI rollout change
12.3Feedback loops
12.4Uptake & cultural fit
12.5Adoption → value

★ Load-bearing gates — the four sub-dimensions named in the W1 meta-essay: strategic posture, decision rights, reversibility, notification triggers.

Three pillars

Strategy

Executive sponsorship, posture, value, resilience

Responsible & Agentic AI Governance

Decision rights, agentic autonomy, risk, observability, compliance

Enablement

Workforce capability, adoption, change management