Generating the Future
summary
Argues that RAG (retrieval-augmented generation) plus embeddings and a vector store solves hallucinations, knowledge‑cutoff and short context by retrieving verified records before calling an LLM; demos load every Oscar winner into a DB and a Langflow pipeline that returns correct 2024 winners without fine‑tuning. Shows a Movies++ prototype that generates interactive UI (posters, trailers, maps) via React Server Components and a forgot‑password flow produced deterministically by function‑call tools (Vercel AI SDK / Ollama / OpenAI), claiming production‑grade UX without retraining models.
AI-written summary · about
chapters (9)
- 0:00Introduction and context
- 2:29Defining AI engineering
- 5:55Understanding AI technology
- 9:42Challenges in generative AI
- 13:00Implementing RAG techniques
- 22:48Generative UI concepts
- 28:12Technical implementation
- 32:08Conclusion and live demo
- 36:04Q&A session
related talks
48mGoing Beyond Prompt Engineering
35mBoost LLMs with Event Streaming and Retrieval Augmented Generation
01h 33mRAG How To Talk To Your Data
50mRetrieval Augmented Generation Beyond Toy Demos
31mRetrieval Augmented Generation
27mAI-Generated React Server Components
42mTo RAG or not to RAG
27mUnlocking the Potential of Large Language Models