Building a Smarter AI Agent with Neural RAG
summary
Live code builds two agent apps that retrieve from the web via Exa: one using keyword RAG (classic inverted-index search) and one using neural RAG via vector search (embeddings + vector index). Evaluation reports task performance, relevance, and latency: neural retrieval wins on semantic and multi-step queries like “personal sites of engineers in San Francisco who like information retrieval,” while keyword search still serves literal GitHub lookups; demos return 10 results by default and scale to 100 or 1000 results on higher tiers.
AI-written summary · about
chapters (10)
- 0:00The history of web search
- 1:16Redesigning search for AI
- 2:29The technology behind Exa
- 3:22Launching a new search paradigm
- 5:30Why LLMs need external search
- 7:10Human search vs AI search
- 9:03Optimizing for AI capabilities
- 11:44The evolving landscape of queries
- 13:32Exa search API overview
- 15:13Live coding and future outlook
related talks
21mHow RAG is Redefining AI Capabilities
01h 35mTutorial: Build AI-Supercharged RAG Apps with a Vector Database
01h 30mOptimizing Retrieval for Agentic Systems
24mAgentic GraphRAG: AI’s Logical Edge
46mAdvanced RAG with Azure AI Search - top questions answered
33mVector Databases in the Age of AI: Embracing RAG for Enhanced Search Capabilities
23m"Deliver AI at Lightspeed: Streamlined approach to AI strategy"
51mTransforming Banking with Advanced AI: GenAI, RAG, and Multi-Agent Systems