Structuring the Video Internet: How Vpin Uses Neo4j to Turn Watching into Learning
summary
Vpin converts passive video watching into structured, searchable learning with a four-step pipeline: curate intuitively (capture URLs or share links), AI extraction into time-stamped highlights and summaries (uses LLMs and OpenAI Whisper for transcription), Neo4j stores a navigable personal knowledge graph and powers real-time queries. Proof of concept shows 2.5 million user spaces, 15,000 curators and 20,000 VPins, and Neo4j handles millions of transactions to enable retrieval-augmented, on-demand one-page summaries.
AI-written summary · about
chapters (8)
- 0:00Introduction
- 1:16Video as learning challenge
- 3:52Four-step learning process
- 8:29How Vpin works
- 15:52Neo4j in the tech stack
- 21:05Personal knowledge graphs
- 22:31Structuring the video internet
- 23:16Q&A session
related talks
23mConnecting the Dots With Context Graphs
01h 00mGraph Mastery to AI
31mBeyond Vectors: Evolving GenAI through Transformative Tools and Methods
01h 17mGraph Database and Retrieval Augmented Generation
22mAI-Powered Personalized Discovery with Graph Search and VLM-Enhanced Embeddings
30mMaking Virtuous Cycles: Lessons From the Frontier of LLM-Knowledge Graph Interaction
30mNeurosymbolic Architecture in Production with Neo4j, GDS, Node2Vec & LLM Grounding
30mPerformant GraphRAG Improving Neo4j to Drive Autonomous AI ...