What is a RAG agent?

A RAG agent is a retrieval-augmented generation system wired into an agent loop, so it can search multiple times, choose between different sources or tools, and decide for itself when it has enough information to answer — instead of doing one fixed search-then-generate pass.

How is a RAG agent different from plain RAG?

Plain RAG is a single lap: it takes your question, converts it into a search, pulls back some documents, and stuffs them into the prompt for the model to write an answer from. If the first search misses the mark, you just get a worse answer — there's no second try built in. A RAG agent adds a decision-making layer on top.

Before answering, it can check whether its first retrieval actually addressed the question, run a follow-up search with different keywords, query a different source (a database instead of a wiki, say), or break a big question into smaller ones it searches separately.

How does a RAG agent decide when to stop searching?

The "agent" part is a loop built around one repeating question: do I have enough to answer well yet? After each retrieval, the model reviews what came back, judges whether the evidence is sufficient, and then either searches again or writes the final answer. Along the way it can pick the right tool for the job — a database query for structured facts, a document search for explanations — and stitch results from several searches together.

That self-checking is what separates it from a pipeline that blindly returns whatever the first search happened to find.

When should you use a RAG agent instead of plain RAG?

The honest trade-off: all that extra searching and reasoning costs more time and more tokens than a single retrieval pass, so it's overkill for simple lookups like "what's our return policy?" It earns its keep on harder, multi-part questions where the answer isn't sitting in one document — research questions, multi-hop questions, or anything that needs comparing several sources before it can be answered well.