AI RESEARCH
Retrieval as Reasoning: Self-Evolving Agent-Native Retrieval via LLM-Wiki
arXiv CS.CL
•
ArXi:2605.25480v1 Announce Type: new LLM agents require retrieval to behave less like one-shot context fetching and like reasoning: searching, reading, traversing, and deciding when evidence is sufficient. However, Retrieval-Augmented Generation (RAG) typically organizes external knowledge as flat chunks retrieved by embedding similarity, exposing a retrieval-as-lookup interface that is poorly aligned with tool-using agents.