# Farzapedia Farzapedia is a personal Wikipedia built by Farza (FarzaTV) using an LLM to process 2,500 entries from his diary, Apple Notes, and iMessage conversations. The system generated 400 detailed articles covering friends, startups, research areas, and personal interests, all interconnected with backlinks. ## Key Insight: Built for the Agent The most distinctive aspect of Farzapedia is that the wiki was not built for the human. It was built for the agent. The file-system structure with backlinks is easily crawlable by any AI agent, making it a more effective knowledge base than [[Retrieval-Augmented Generation (RAG)]]-based approaches. An agent can start at `index.md` and drill into specific pages as needed. ## How It Works Farza can spin up Claude Code on the wiki and ask cross-domain questions like asking for landing page inspiration. The agent reads wiki articles spanning philosophy notes from a documentary, competitor analysis with screenshots, and saved visual inspiration from years ago, then synthesizes a useful answer. As new content is added (articles, images, meeting notes), the system automatically updates 2-3 existing articles where the context belongs, or creates a new article. Farza describes it as a genius librarian that never gets tired. ## Compared to RAG Farza built a similar system a year earlier using RAG, but found it inferior. A knowledge base that lets an agent navigate via a file system it actually understands works better than embedding-based retrieval at this scale. ## Relation to LLM Wiki Farzapedia is an independent implementation of the [[LLM Wiki]] pattern described by [[Andrej Karpathy]]. Where Karpathy focuses on research knowledge bases, Farzapedia demonstrates the pattern applied to personal life documentation. ## References - https://x.com/FarzaTV/status/2040563939797504467 ## Related - [[LLM Wiki]] - [[Andrej Karpathy]] - [[Retrieval-Augmented Generation (RAG)]] - [[Personal Knowledge Management (PKM)]] - [[Knowledge Graph (KG)]] - [[Agentic Knowledge Management (AKM)]] - [[Compounding Knowledge]]