In the context of Knowledge Management (PKM) and AI-enhanced workflows, “Gateway” and “Pillar” notes—often referred to as MOCs (Maps of Content)—are organizational structures designed to bridge the gap between fragmented information and a cohesive knowledge base. They act as “anchor points” that simplify the retrieval of context for LLMs by providing a centralized index of a specific domain or theme.
General Concepts: Gateway vs. Pillar Notes
-
Gateway Notes: These function as high-level navigational hubs. Think of them as the “Table of Contents” for a broad topic (e.g., “Engineering Principles”). They do not necessarily contain deep technical detail; rather, they serve as a directory that directs a user (or an AI) to the most important sub-topics or specific notes.
-
Pillar Notes: These are more substantive. They act as a foundational reference for a specific subject (e.g., “Chemical Thermodynamics”). They contain core concepts, definitions, and theories. An AI uses these to understand the “source of truth” within your vault, preventing it from getting lost in dated or secondary research.
Why They Improve AI Retrieval
When you ask an AI to synthesize information from a large Obsidian vault, it can struggle to determine which notes are authoritative. By creating these structures:
-
Reduced Noise: You create a curated pathway, ensuring the AI focuses on your primary sources rather than draft or fleeting notes.
-
Contextual Anchoring: The AI can perform a “breadth-first” search (starting at the Gateway) followed by a “depth-first” search (accessing the Pillar), mimicking a logical research process.
-
Reduced Hallucination: When the AI is “grounded” in an MOC, it is statistically more likely to pull relevant, verified information from your linked notes instead of relying solely on its internal training data. - on AI
🧠 Why this version works better
- Clear sections → easier for AI chunking
- Explicit concepts → matches your future queries
- Link-rich → creates strong retrieval signals
- Readable → works for you, not just the model
🔥 About your embed idea (you’re thinking ahead)
You’re exactly right — the next evolution is:
turning this into a “live dashboard”
In Obsidian, that means things like:
-
embedding key notes:
![[ai - opportunity or threat]] -
or even specific sections:
![[HITL - AI enabled service delivery#summary|Summary]]
🧩 What that unlocks
- this note becomes self-updating context
- Copilot sees both:
- the summary ✅
- the source material ✅
👉 massively improves answers
⚡ One pro tip (worth it early)
Add a small section to key notes like:
## Summary
Then embed just that.
🚀 Where you’re heading (and it’s a good place)
You’re building:
a queryable knowledge system with layered abstraction
- raw notes → detail
- structured notes → thinking
- gateway notes → access
If you want next step, we can:
- design a standard “summary block” format for all your notes
- or set up a pattern for cross-vault ingestion (Logseq → Obsidian → AI)
That’s where this really starts compounding.