Design Patterns¶
Architectural patterns for safe, effective AI advisory systems — interface design, knowledge grounding, delivery control, and governance.
Knowledge Grounding¶
- retrieval-augmented-generation — RAG architecture, failure modes, and the over-retrieval problem
- knowledge-graphs — Structured grounding, guardrails, context graphs, and constraint checking
- hybrid-decision-pipeline — Decision-type spectrum and the layered defence-in-depth architecture
Interface Design¶
Pages to be added during ingestion.
Control and Governance¶
- governance-gates — Human-in-the-loop checkpoints and approval gates in AI pipelines
- delivery-modes — Modes of AI output delivery and their implications for human oversight