What is agentic RAG?
Agentic RAG is a way of building AI assistants that plan their own searches, retrieve information from several sources, check what they found and only then write an answer with references. Basic RAG runs one search and answers from it. Agentic RAG suits questions that need several documents or systems.
How does basic RAG work?
Retrieval-augmented generation finds relevant passages in your documents and gives them to the AI model, so the answer comes from your material instead of the model's memory.
What does the agentic part add?
The assistant breaks a question into steps, decides where to look for each step, compares the results and retries when something is missing. That makes it better at questions that span several sources.
When is it worth it?
When questions regularly need information from more than one place, such as policies plus records plus a database. For simple questions about one set of documents, basic RAG is enough and costs less.
