The RAG Really Ties the App Together
summary
Explains building a RAG chat app using Elasticsearch and Elastic Playground, deploying an inference API endpoint for embeddings and Azure OpenAI for completions, automatic chunking (250-word chunks with 100-word overlap), and exporting working Python FastAPI + React code. Demonstrates ELSER sparse vectors with token expansion and shows token pruning can speed search 3x–7x with little recall loss. Implements generative caching for semantically similar question reuse; demo functions with minor caveats around streaming and prompt tuning.
AI-written summary · about
chapters (12)
- 0:00Intro
- 1:44Lexical 101
- 3:28Semantical 101
- 9:20ELSER
- 12:16Query problem
- 15:05Inference API
- 16:27Semantic text
- 23:17Generative caching
- 26:06LLM caching benefits
- 27:19LLM helpy helper
- 31:19Demo
- 43:20Outro
related talks
41mFrom RAGs to riches: Build an AI document interrogation app in 30 mins
42mTake Your RAG To The Max!
27mFrom RAGs to riches: Build an AI document interrogation app in 30 mins
48mGoing Beyond Prompt Engineering
16mBuild a RAG app in minutes using Langflow OpenAI and Azure
50mRetrieval Augmented Generation Beyond Toy Demos
01h 36mTutorial: From Planni...
45mMaking semantic search & RAG real: How to build a prod-ready app