DeepYardDeepYard
A

Autonomous RAG

Self-managing RAG system that decides when and how to retrieve information

Open SourceFree

About

An autonomous RAG implementation where the agent decides whether retrieval is needed, what to search for, which retrieval strategy to use (vector, keyword, or hybrid), and how many rounds of retrieval to perform. Unlike static RAG pipelines, this system adapts its retrieval behavior based on the question complexity and initial results quality. Part of the awesome-llm-apps collection.

Details

Patternagent-loop
StepsAnalyze query to determine retrieval need → Select optimal retrieval strategy → Execute retrieval and evaluate results → Iterate or generate final response
Use CasesComplex Q&A systems, Adaptive information retrieval, Dynamic knowledge bases

Tags

ragautonomousadaptiveworkflow