RAG is dead, right?? — Kuba Rogut, Turbopuffer

By ai.engineer

Categories: AI, Tools

Summary

RAG isn't dead—it's evolving. Hybrid retrieval combining vector search, full-text search, and agentic tools is becoming the standard for serious applications, with Cursor seeing 13-24% accuracy gains from semantic search integration in their code indexing system.

Key Takeaways

  1. RAG misconceptions: Most people think RAG is just vector search, but it encompasses vector search, full-text search (BM25), regex filtering, and basic filters—all feeding into LLM generation.
  2. Agentic search isn't file system grep alone. It's giving agents tools to iteratively find and reason over context, progressively refining searches until task completion—like Cursor's multi-step code navigation.
  3. Cursor's Merkle tree optimization: Use cryptographic hash trees to detect similar codebases across teams, then only re-chunk/re-embed changed files instead of entire repos—reducing embedding costs dramatically.
  4. Semantic search ROI is quantifiable: Cursor measured 12.5-13.5% average accuracy improvement and 24% improvement for Composer model through integrated semantic search in their internal benchmark.
  5. Google search volume data validates RAG evolution: Despite 'RAG is dead' tweets, search volume for RAG hit a new inflection point mid-2025, indicating sustained enterprise adoption beyond hype cycles.

Related topics

Transcript Excerpt

[music] >> All right, welcome everyone. Thanks for coming out. I see it's a full room, so I appreciate everyone coming out. So welcome to the talk about rag is dead, right? So my name is Kuba. I'm deployed engineer at Turbo puffer. So for those that don't know what Turbo puffer is, we are a full text search and vector search database built from first principles on top of object storage. If you would love to learn more, I'll just come find me after the talk if you have any questions. So let's get started. So this talk I get is up sorry about how rag is dead, how hybrid tool tool rich retrieval is becoming a default for serious agentic search. So you guys have been on Twitter or other social media platforms or I guess X they call it now, you might have seen a lot of tweets like this about ho…

More from ai.engineer