DATA RETRIEVAL FROM VARIOUS SOURCES USING RETRIEVAL AUGMENTED GENERATION AND MULTI-AGENT WORKFLOW: A COMPARATIVE ANALYSIS

The emergence of Large Language Models has revolutionized artificial intelligence applications, yet fundamental limitations including knowledge currency and hallucination phenomena constrain their practical deployment.This research presents a comprehensive comparative analysis of Retrieval-Augmented Generation systems and Multi-Agent architectures implemented within Azure Durable Functions serverless framework.We developed specialized agents for SQL database operations, data visualization, and web content extraction, coordinated through an intelligent manager agent.Experimental evaluation demonstrates that Multi-Agent systems achieve thirty-seven percent reduction in response latency compared to traditional microservice architectures while improving query resolution accuracy by forty-two percent.RAG implementations achieved eighty-nine percent accuracy in factual question-answering tasks with ninety-four percent verifiable source attribution.The research establishes comprehensive frameworks for architectural selection based on empirical performance analysis across diverse use cases.Our hybrid approach combining RAG-enhanced agents demonstrates twenty-three percent performance improvement over pure Multi-Agent systems for knowledgeintensive tasks.The findings provide systematic guidance for organizations implementing intelligent AI systems in cloud-native environments.