Querying Live Data With LLM App
summary
Claims a working pipeline that builds an AI app in 30 lines of code using OpenAI API and Pathway Python libraries, ingests Dropbox and streaming sources, does realtime embedding and indexing without an external vector DB, and pushes alerts to Slack. Demo shows a discounts app and Dropbox Q&A running in Docker, session replay caching for offline tests, conversion of batch code to streaming with one-line change, ~1k GitHub stars and two open-source users running it in production.
AI-written summary · about
chapters (10)
- 0:00Introduction
- 1:09Why LLMs matter today
- 6:08LLM API and RAG limits
- 10:04RAG architecture challenges
- 13:19Introducing LLM App
- 18:37LLM App features
- 20:46Real-time data with LLM App
- 23:43Demo: Dropbox expense summarization
- 29:48Demo: Flight status notifications
- 34:42Summary and next steps
related talks
29mBuild AI-powered data pipeline without vector databases
24mDemocratizing data access through LLM
52mBuild an Intelligent Python App, No Infra Hassles!
47mBuilding Python based AI Systems with LLMs
28mAPIs and GenAI : Your Perfect Combo with LLM-Powered Frameworks.
26mWelcome to the AI Jungle! Now What?
40mRolling out Enterprise AI: Tools, Insights, & Team Empowerment
40mOptimizing AI Applications for Production with Observability