Why LLM Wiki? Future Of Knowledge For Agentic AI & Humans

This video explains building personal and agentic knowledge graphs (LLM Wiki) to structure information for human and AI use. Callum demonstrates nodes, edges, triples, scaling a personal Obsidian graph, and how graph-based RAG improves AI retrieval. He contrasts human and agent vaults and outlines an automated LLM Wiki workflow for ingestion, integration, and maintenance.

• Core model: Nodes, edges and triples are the atomic units that form a scalable knowledge graph.
• Practical build: Live Obsidian demo shows how linking notes creates a compounding personal graph over years.
• AI integration: Standard RAG works for single-document answers; graph RAG outperforms it for reasoning across connected sources.
• LLM Wiki workflow: Layered approach—raw sources, AI-compiled interlinked wiki, and agent-driven maintenance keeps the shared knowledge current and usable by multiple tools.

Quotes:

Google built one for the entire internet. I built one for my own brain.

A triple is the atom of a knowledge graph: subject, relationship, and object.

The LLM reads sources, extracts key information, and integrates it into a persistent wiki – knowledge compiled once and kept current.