Event Driven AI: Supercharging ChatGPT with RAG & LangStream
summary
Argues that combining RAG with event streaming lets ChatGPT-style models answer from live data by wiring retrieval and generation into pipelines. Uses LangStream plus Kafka or Apache Pulsar/Kafka sources, vector stores like Pinecone/Milvus/Cassandra or PGVector, and OpenAI (demo runs on OpenAI GPT‑3.5) to crawl sites, chunk text, compute embeddings, write to a vector DB, then stream query results into a chat completion. Demo shows a runnable YAML pipeline and a working RAG chatbot deployed via Docker/Kubernetes.
AI-written summary · about
chapters (9)
- 0:00Introduction and agenda
- 2:05Basics of generative AI and ChatGPT
- 10:14Limitations of large language models
- 20:48Improving LLMs with RAG
- 27:31RAG framework explained
- 32:03Introducing LangStream
- 38:40LangStream pipelines and demos
- 41:29Live demo setup
- 48:41Resources and closing
related talks
27mBuild a ChatGPT RAG Data Pipeline
58mDemystifying GenAI: Build a ChatGPT App with Vector Store
35mBoost LLMs with Event Streaming and Retrieval Augmented Generation
49mBuild ChatGPT RAG Data Pipeline with RisingWave Stream Processor & Vector Store
25mLets Build A Node.js ChatGPT-Like Bot With AI APIs
30mGenAI Beyond Chat with RAG, Knowledge Graphs, and Python
41mFrom RAGs to riches: Build an AI document interrogation app in 30 mins
48mAn Introduction to Retrieval Augmented Generation