Creating a Board Game Chatbot with Postgres, AI, and RAG
Friday, October 25 at 13:40–14:30
This session explores the integration of PostgreSQL with AI technologies and Large Language Models (LLMs), demonstrating the creation of a chatbot capable of answering board game rules questions. Using board games as a fun example, we'll showcase information retrieval techniques based on natural language and how to implement them with PostgreSQL at scale.
We'll start with an introduction to Retrieval-Augmented Generation (RAG), a cutting-edge approach that combines retrieval-based and generation-based models for accurate, contextually relevant responses. The session will cover the process of collecting and preprocessing board game rules data, generating and storing vector embeddings using pgvector for efficient similarity searches, and training and deploying a language model to generate responses using these embeddings. Additionally, we'll discuss configuring PostgreSQL for highly performant vector searches and real-time data operations.
Attendees will gain insights into RAG's theoretical underpinnings and practical skills in integrating AI with PostgreSQL for intelligent chatbot applications. Join us to discover how combining PostgreSQL with advanced AI techniques and LLMs can revolutionize data interaction.
