Agentic Retrieval-Augmented Generation (Agentic RAG) is an extension of retrieval-augmented generation that introduces agents to achieve a more flexible retrieval and decision-making process.

The Agent-Enhanced Retrieval Generation (AEG) approach, developed between 2024 and 2025, is a system paradigm that has gradually evolved through the integration of RAG and agent technologies. This method adds capabilities such as task planning, information routing, and result verification to the traditional retrieval generation framework based on large language models.

In Agentic RAG, agents not only perform simple information retrieval, but also dynamically select data sources according to task requirements, distribute queries to different knowledge bases or tools, and filter, verify and integrate the retrieval results to generate more accurate output.

This method is widely used in complex question-answering systems, professional knowledge retrieval, and automated decision-making scenarios, and represents an important development direction for the integration of intelligent agent systems and retrieval technologies.