Obsidian Knowledge System

The Obsidian Knowledge System is the foundation layer of the entire 1:N stack: years of accumulated knowledge made instantly searchable through vector embeddings.

The Scale

• 2,353 markdown documents
• 32,163 vector chunks indexed
• Semantic search via MCP integration
• Daily indexing on DGX Spark

What's In There

• Research paper analyses and summaries
• Project documentation and decisions
• Meeting notes and action items
• Technical learnings and experiments
• Personal reflections and insights
• Code snippets and solutions

The Vector Pipeline

Every night at 3 AM, DGX Spark indexes new and modified documents:

1. Extract text from markdown files
2. Chunk into semantic units
3. Generate embeddings with local models
4. Store in LanceDB for fast retrieval
5. Expose via MCP for Claude access

Why It Matters

Most knowledge management is write-only: you capture information but rarely retrieve it effectively. Vector search changes this equation. Any question can surface relevant context from years of accumulated knowledge.

The Foundation Effect

Everything builds on this layer:

• Continuum queries it for personal context
• AI Explorer draws from research notes
• Axon uses it for meeting intelligence
• Blog posts reference accumulated learnings

The Compound Interest of Knowledge

Every document added increases the value of semantic search. Cross-connections emerge automatically. Old insights surface when relevant to new questions. The vault becomes smarter over time.