Why the Best AI Coding Tools Abandoned RAG (And What They Use Instead)

Claude Code, Cline, Aider, Codex CLI - every top AI coding tool has abandoned traditional RAG and embeddings. But for enterprise knowledge bases, RAG is a $2B market growing 50% per year, and it's ~100x cheaper per query than full context loading. So is RAG dead or not? Both sides are right - they're just talking about different problems. Code is structured and perfectly searchable with grep. Business documents aren't. In this video I break down where the line is and what to use on each side. ~~~~~~~~~~~~~~~~~~~~~~~~~~ The best way to learn to code is with Scrimba - start free and get 20% off the Pro plan: - Fullstack Developer Path (featured in this video): https://scrimba.com/fullstack-path-c0fullstack?via=colemedin - All Scrimba coding courses (many free): https://scrimba.com/courses?via=colemedin ~~~~~~~~~~~~~~~~~~~~~~~~~~ - The Dynamous Agentic Coding Course is now FULLY released - learn how to build reliable and repeatable systems for AI coding: https://dynamous.ai/agentic-coding-course ~~~~~~~~~~~~~~~~~~~~~~~~~~ Sources: - Boris Cherny / Claude Code on agentic search: https://smartscope.blog/en/ai-development/practices/rag-debate-agentic-search-code-exploration/ - Why Cline doesn't use RAG for AI coding: https://pashpashpash.substack.com/p/why-i-no-longer-recommend-rag-for - LlamaIndex - Agentic Retrieval: https://www.llamaindex.ai/blog/rag-is-dead-long-live-agentic-retrieval - Aider repo map (Tree-sitter approach): https://aider.chat/docs/repomap.html ~~~~~~~~~~~~~~~~~~~~~~~~~~ 0:00 Is RAG Dead? 0:55 All Search is RAG (but not Semantic Search) 1:52 Structured vs. Unstructured Data 4:09 Why RAG Fails for Code 6:22 Scrimba 7:29 Agentic Search in Practice 11:05 When to Use Traditional RAG 13:12 Bridging Both Retrieval Strategies 14:35 Final Thoughts ~~~~~~~~~~~~~~~~~~~~~~~~~~ Join me as I push the limits of what is possible with AI. I'll be uploading videos weekly - at least every Wednesday at 7:00 PM CDT!