Keyword search is all you need: Achieving RAG-Level Performance without vector databases using agentic tool use
#Retrieval‑Augmented Generation #vector database #keyword search #agentic framework #tool‑augmented LLM #information retrieval #semantic search #cost‑effective #knowledge base updates #arXiv #cs.IR #performance metrics
📌 Key Takeaways
- RAG traditionally depends on vector databases for contextual retrieval.
- Agentic, tool‑augmented LLMs provide alternative retrieval mechanisms.
- The study compares standard RAG systems with agents that use basic keyword search.
- Empirical results show keyword‑search agents achieve over 90% of traditional RAG metrics.
- The approach is simple, cost‑effective, and ideal for frequently updated knowledge bases.
📖 Full Retelling
In a study released on arXiv (v1, 19 Dec 2025) by Shreyas Subramanian, Adewale Akinfaderin, Yanyan Zhang, Ishan Singh, Mani Khanuja, Sandeep Singh, and Maira Ladeira Tanke, the authors investigate whether Retrieval-Augmented Generation (RAG) performance can be achieved without vector databases by leveraging agentic keyword search. Published in the field of Information Retrieval (cs.IR), the paper argues that vector databases and semantic search add little value over simple keyword search when integrated into an agentic framework, demonstrating that such setups can reach over 90% of traditional RAG performance while reducing complexity, cost, and enabling faster knowledge‑base updates.
🏷️ Themes
Retrieval‑Augmented Generation (RAG), Agentic AI and tool‑augmented LLMs, Information retrieval methods, Cost and complexity of AI systems, Vector vs. keyword‑based search, Scalable knowledge‑base management
Entity Intersection Graph
No entity connections available yet for this article.
Original Source
--> Computer Science > Information Retrieval arXiv:2602.23368 [Submitted on 19 Dec 2025] Title: Keyword search is all you need: Achieving RAG-Level Performance without vector databases using agentic tool use Authors: Shreyas Subramanian , Adewale Akinfaderin , Yanyan Zhang , Ishan Singh , Mani Khanuja , Sandeep Singh , Maira Ladeira Tanke View a PDF of the paper titled Keyword search is all you need: Achieving RAG-Level Performance without vector databases using agentic tool use, by Shreyas Subramanian and 6 other authors View PDF HTML Abstract: While Retrieval-Augmented Generation has proven effective for generating accurate, context-based responses based on existing knowledge bases, it presents several challenges including retrieval quality dependencies, integration complexity and cost. Recent advances in agentic-RAG and tool-augmented LLM architectures have introduced alternative approaches to information retrieval and processing. We question how much additional value vector databases and semantic search bring to RAG over simple, agentic keyword search in documents for question-answering. In this study, we conducted a systematic comparison between RAG-based systems and tool-augmented LLM agents, specifically evaluating their retrieval mechanisms and response quality when the agent only has access to basic keyword search tools. Our empirical analysis demonstrates that tool-based keyword search implementations within an agentic framework can attain over $90\%$ of the performance metrics compared to traditional RAG systems without using a standing vector database. Our approach is simple to implement, cost effective, and is particularly useful in scenarios requiring frequent updates to knowledge bases. Subjects: Information Retrieval (cs.IR) ; Artificial Intelligence (cs.AI) Cite as: arXiv:2602.23368 [cs.IR] (or arXiv:2602.23368v1 [cs.IR] for this version) https://doi.org/10.48550/arXiv.2602.23368 Focus to learn more arXiv-issued DOI via DataCite Submission history F...
Read full article at source