Agent vs LLM vs RAG (Mental Model)

A simple analogy clarifies the relationship between these three commonly conflated terms: the LLM is the brain, RAG is feeding that brain with fresh information, and an Agent is the decision-maker that plans and acts using the brain and the available tools.

LLM (Large Language Model): trained on massive text data, an LLM like GPT-4 can reason, generate, and summarize — but only using what it already knows from its training data. It's smart, but static: it can't access the web, call APIs, or fetch new facts on its own.

RAG (Retrieval-Augmented Generation): enhances an LLM by retrieving external documents (from a vector database, search engine, etc.) and feeding them into the LLM as context before generating a response. RAG makes the LLM aware of updated, relevant information without retraining it.

Agent: adds autonomy to the mix. An agent doesn't just answer a question — it decides what steps to take: should it call a tool? Search the web? Summarize? Store information? An agent uses an LLM, calls tools, makes decisions, and orchestrates workflows, much like a real human assistant would.

In short: the LLM provides raw reasoning and generation capability; RAG extends what the LLM knows; and an Agent decides how and when to use both — the LLM's reasoning and RAG's fresh information — to actually accomplish a multi-step goal.