Beyond Vectors: Evolving GenAI through Transformative Tools and Methods
summary
Built an LLM app called Agent Neo that grounds answers in Neo4j by indexing docs as 512‑char chunks with embeddings (GCP) and retrieving with KNN (cosine similarity); LangChain + Streamlit serve the UI and every prompt plus its context documents is logged into the graph. Community detection, centrality and PageRank expose redundant clusters, let node_collapse preserve links while deduplicating, and reveal long‑tail conversation lengths; code and demos are open source on GitHub and the system demonstrably improves grounding and debugs missing data.
AI-written summary · about
chapters (16)
- 0:00Graph RAG introduction
- 1:10What is retrieval augmented generation
- 3:39Graph databases role
- 5:50Agent Neo purpose
- 7:13High-quality grounding data elements
- 8:37Exploratory data analysis with graphs
- 12:51Document similarity analysis
- 17:48Logging conversations for explainability
- 22:45Conversation flow tracking
- 26:22Improving RAG with logs
- 31:18Response patterns analysis
- 34:09User satisfaction evaluation
- 36:17Scalable RAG systems
- 37:02SEC data RAG demo
- 40:19Agent Neo grounding demo
- 41:03Conclusion and resources
related talks
27mWhen Vectors Aren’t Enough: Unlocking AI’s Hidden Connections with Graphs
38mKnowledge Graphs : Adding a New Dimension to Vector Search and Generative AI
58mAI-Enhanced Big Data: Integrating Private LLMs and Vector Databases
01h 41mGraph Intelligence: Enhance Reasoning and Retrieval Using Graph Analytics
46mEvent Streaming and its Potential for the GenAI World
57mEnter The Brave New World of GenAI with Vector Search
48mEnter the Brave New World of GenAI with Vector Search
01h 19mIntro to GraphRAG