VKL-004 — VISUAL KNOWLEDGE LIBRARY
Version 1.1.0
Answer: How can this be applied?

├── V-THINK — AI as a Thinking Partner
│   Answers: What changes when AI is used as a governed thinking partner?
│   Application: Orientation
│   Governing pages: PAGE-HOME, PAGE-CAP
│   Reliance: Use to explain the division of labour between humans and AI.
│   Starter prompt: Explain this visual to a beginner, identify one failure mode and connect it to Permission-to-Rely.
├── V-SKILLS — GovAIaaS Specialist AI Skills
│   Answers: Which specialist capabilities can be combined?
│   Application: Capability selection
│   Governing pages: PAGE-CAP, PAGE-HOW
│   Reliance: Use to select or combine specialist capabilities. The original visual label AI Approve is superseded by Permission-to-Rely Assessment.
│   Starter prompt: Map my task to the minimum skill combination and state which human approvals remain outside AI authority.
├── V-MECH — Candidate Mechanisms Innovation
│   Answers: How can AI help generate and test candidate mechanisms?
│   Application: Mechanism discovery
│   Governing pages: PAGE-INNOV, PAGE-HOW
│   Reliance: Use to move from evidence collection toward multiple testable causal explanations.
│   Starter prompt: Generate a mechanism-discovery workflow, alternatives and discriminating tests without claiming confirmation.
├── V-CLAIMS — One Climate Claim, Many Claims
│   Answers: What component claims are hidden inside one climate statement?
│   Application: Claim decomposition
│   Governing pages: PAGE-CLIMATE
│   Reliance: Use to expose the sub-claims hidden inside a compressed climate statement.
│   Starter prompt: Decompose the claim, classify each component and identify the weakest necessary component.
├── V-4AI — One Climate Claim, Four AIs, One Audit Template
│   Answers: How can multiple AI audits be compared using one template?
│   Application: Multi-AI comparison
│   Governing pages: PAGE-CLIMATE, PAGE-CASE
│   Reliance: Use to compare independent outputs under one audit template. Convergence is process evidence, not proof.
│   Starter prompt: Compare four AI audits, separate convergence from proof and identify repeated weak links.