Connected Data on X: "Graphify: An open-source tool that builds upon Andrej Karpathy's personal knowledge base concept to achieve fully multimodal automated graph generation Andrej Karpathy's LLM Wiki got viral.  Three layers: Raw sources: your curated, immutable document collection The wiki: https:

Graphify: An open-source tool that builds upon Andrej Karpathy's personal knowledge base concept to achieve fully multimodal automated graph generation Andrej Karpathy's LLM Wiki got viral.  Three layers: Raw sources: your curated, immutable document collection The wiki: LLM-generated, interlinked markdown pages The schema: a CLAUDE.md / AGENTS.md config that defines structure and workflows Three operations: Ingest: drop a source, the LLM processes it, updates 10-15 wiki pages Query: ask questions, file good answers back as new pages Lint: periodic health-checks for contradictions, orphan pages, stale claims Graphify addresses the pain points of Karpathy's original workflow: tedious manual organization, high token consumption, and a lack of tool encapsulation. Graphify supports fully multimodal automated parsing (covering code, documents, and images), utilizes tree-sitter for local AST parsing to save tokens, and employs the Leiden community detection algorithm to build knowledge graphs without vector databases.  Tests show its token consumption is only 1/71.5 of the original method. Additionally, it supports file watching and Git hooks, enabling automated incremental updates for the knowledge base, providing an out-of-the-box 'ultimate' solution for personal knowledge management. Main Points 1. Fully multimodal automated processing, eliminating the burden of manual organization. Built-in multimodal pipelines automatically parse code structures, document semantics, and image concepts, allowing for unified graph integration without manual preprocessing, making everything graphable. 2. Two-stage process significantly optimizes token consumption. Combines local AST parsing with parallel sub-agent semantic extraction, paired with a SHA256 caching mechanism, achieving 71.5x token savings in mixed corpus scenarios. 3. No vector database required, lowering deployment barriers. Clustering is based on graph topology and the Leiden algorithm, eliminating the need for embeddings and complex configurations, with support for one-click generation of interactive HTML graphs. 4. Dynamic monitoring and incremental updates to maintain knowledge relevance. Supports file watching mode and Git hooks, automatically triggering parsing when content changes, ensuring the knowledge base grows continuously with new data without needing to rebuild the entire graph. By Safi Shamsi Github github.com/safishamsi/gra… #LLM #EmergingTech #GraphRAG #PKM #AIAgents #DataEngineering #DataModeling -- Connected Data London 2026 has been announced! 11-12 November, Leonardo Royal Hotel London Tower Bridge 📝 connected-data.london/post/cdl-2026-… Join us for all things #KnowledgeGraph #Graph #analytics #datascience #AI #graphDB #SemTech #Ontology 🎟 Ticket sales are open. Benefit from early bird prices with discounts up to 30%. 2026.connected-data.london/?utm_source=tw… 📺 Sponsorship opportunities are available. Maximize your exposure with early onboarding. Contact us at [email protected] for more.