In this talk, we will explore how the RAG design pattern enables the integration of Generative AI applications with custom data sources using Neo4j's graph database. You will learn how to augment large language models (LLMs) by adding specific and contextual data from structured and unstructured sources, creating more reliable and accurate responses in generative AI systems.

Key aspects of the talk include:

• Understanding the core concepts of Retrieval Augmented Generation (RAG) and its practical applications.
• How to combine structured data (from databases) and unstructured data (like text files and PDFs) using a graph-based knowledge stack.
• Demonstrating Neo4j’s capabilities in supporting vector search, graph queries, and data modeling strategies for scalable Gen AI solutions.
• Multiple GraphRAG Data models

Key Takeaways

• Practical understanding of how to integrate graph databases with LLMs using RAG.
• Best practices for creating a minimum viable graph (MVG) for generative AI applications.
• Insights into leveraging Neo4j’s Gen AI ecosystem for faster development of accurate and reliable Gen AI applications

Target Audience

This talk is tailored for professionals with varying levels of familiarity with Neo4j, including:

• Beginner to Intermediate Neo4j Users: Individuals who are already working with Neo4j or other graph databases and want to deepen their knowledge in integrating AI and knowledge graphs.
• AI and Data Science Practitioners: Those interested in how graph databases can enhance AI models, particularly in improving context and accuracy for Generative AI applications.
• Solution Architects & Data Engineers: Professionals looking to leverage Neo4j in their data engineering pipelines, especially in combining structured and unstructured data sources for enterprise AI use cases.

Pre-requisites

• Basic knowledge of Neo4j and graph database concepts.
• Some familiarity with AI and LLMs (e.g., ChatGPT) is beneficial but not required.