From RAG to Agents: Rearranging Search Building Blocks for Better Retrieval
hugobowne · x · 2026-08-31
Classic RAG offers a single retrieval chance; failure means a bad answer. Agents can inspect results, reject them, reformulate queries, and re-search using BM25, embeddings, SQL, or grep. Doug Turnbull (ex-Reddit, Shopify, Wikipedia) explains how this rearranges search building blocks in his Context Camp lightning lesson. Full recording and annotated slides are available.
More from coding & agent
- CodeKeep Workflow: Using Agents to Auto-Collect GitHub Code Snippets — JordanMorgan10 · 2026-08-31
- Luca Agent Tracks All Your Coding Sessions Across Platforms — RileyRalmuto · 2026-08-31
- Vibe coders are getting sued: a pre-launch security checklist from 60+ shipped MVPs — PrajwalTomar_ · 2026-08-31
- RSI-Exam released: Benchmarking recursive self-improvement in AI agents — HuaxiuYaoML · 2026-08-31
- 818 Open-Source Cybersecurity Skills Empower AI Agents Across 6 Frameworks — tom_doerr · 2026-08-31
- Using Grok bot to handle web scraping tasks without APIs — PTrubey · 2026-08-31