# What is agentic RAG? > Agentic RAG gives an AI agent document search as a tool, so the agent decides when and what to look up instead of searching before every reply. Source: https://learn.ynteractive.com/content/glossary/agentic-rag · Updated 2026-10-21 · Free from Stepthrough (https://learn.ynteractive.com) [AI agents](https://learn.ynteractive.com/content/glossary#topic-ai-agents) · [AI glossary](https://learn.ynteractive.com/content/glossary) Quick answer Agentic RAG is retrieval-augmented generation where an AI agent decides when to search your documents and what to search for. The document search is one of the agent's tools, like email or a calendar. The agent can search, read the results, search again with better words, or skip searching when it isn't needed. Last updated October 21, 2026 [Take the free module →](https://learn.ynteractive.com/courses/ai-for-customer-support/14)Module 14 of the free [AI for Customer Support](https://learn.ynteractive.com/courses/ai-for-customer-support) course, about 13 minutes. No account needed. Updated October 21, 2026 · By [Robert Breen](https://learn.ynteractive.com/content/about) ## Why it matters for a small business In classic [RAG](https://learn.ynteractive.com/content/glossary/rag), every message triggers one search and the top results are pasted into the prompt. That works, but it searches even for "thanks!" and can't try again when the first search misses. Agentic RAG hands the search to an [AI agent](https://learn.ynteractive.com/content/glossary/ai-agent) as a tool. The agent writes its own search query, can run several searches for a two-part question, and can combine the results with other tools. The trade-off is control. An agent that decides for itself might skip the search and answer from memory, which is exactly what you don't want for policy questions. So agentic RAG setups lean on two things: a clear tool description, so the agent knows what the search covers, and a system message that says when it must search and what to do when nothing is found. ## In a real lesson: Answer Staff Questions from Your Own Docs using n8n + Pinecone In [Answer Staff Questions from Your Own Docs](https://learn.ynteractive.com/content/n8n-ai-agent-pinecone-rag-chatbot), you connect the Pinecone Vector Store to the AI Agent's Tool slot in Retrieve Documents (As Tool for AI Agent) mode. You paste a Description that says it searches the Cedar Ridge handbook for time off, expenses, remote work and more, and the voiceover explains that the agent reads it to decide when to search. Then the system message tightens the choice: always search the handbook before you answer, cite the page, and say "I don't know" when it isn't covered. In the test, the agent searches for vacation days and cites page 6, then searches for the desk chair question and cites page 9. For the bonus question it searches, finds nothing, and says "I don't know". The earlier [support agent](https://learn.ynteractive.com/content/support-agent-answers-from-your-docs-n8n) lesson uses the same agent-with-a-tool pattern, but its Google Docs Tool reads the whole policies document instead of searching a vector store. [Try this lesson free](https://learn.ynteractive.com/modules/pinecone-n8n-rag-chatbot) or [read the step-by-step guide](https://learn.ynteractive.com/content/n8n-ai-agent-pinecone-rag-chatbot). ## Common confusions ### Agentic RAG vs classic RAG Classic RAG always runs one search before the AI answers. Agentic RAG lets the agent choose whether, how often and with what words to search. ### Agentic RAG vs agentic AI Agentic AI is the broad idea of AI that plans and acts. Agentic RAG is one use of it: an agent that retrieves from your documents. ## Tips - Write the tool description like a label on a filing cabinet: what is inside and when to use it. - Tell the agent in the system message to search before every policy answer, so it doesn't answer from memory. - Test a question your documents don't cover and check that the agent says it doesn't know. ## Related terms [Retrieval-augmented generation (RAG)](https://learn.ynteractive.com/content/glossary/rag) · [AI agent](https://learn.ynteractive.com/content/glossary/ai-agent) · [Tool calling](https://learn.ynteractive.com/content/glossary/tool-calling) · [Vector database](https://learn.ynteractive.com/content/glossary/vector-database) · [Agentic AI](https://learn.ynteractive.com/content/glossary/agentic-ai) · [System prompt (system message)](https://learn.ynteractive.com/content/glossary/system-prompt) ## More AI agents terms [AI Agent node](https://learn.ynteractive.com/content/glossary/ai-agent-node) · [AI agent permissions](https://learn.ynteractive.com/content/glossary/agent-permissions) · [AI agent vs assistant](https://learn.ynteractive.com/content/glossary/ai-agent-vs-assistant) · [AI video agent](https://learn.ynteractive.com/content/glossary/ai-video-agent) ## Where you use it: free lessons - [Answer Staff Questions from Your Own Docs using n8n + Pinecone](https://learn.ynteractive.com/content/n8n-ai-agent-pinecone-rag-chatbot) (n8n, 13 min) - [A Support Agent That Answers from Your Own Docs](https://learn.ynteractive.com/content/support-agent-answers-from-your-docs-n8n) (n8n, 13 min) ## Frequently asked questions **Is agentic RAG better than normal RAG?** It is more flexible, especially for multi-part questions, but less predictable. For a simple FAQ bot, a fixed search before every answer can be enough. **How do I make the agent actually use the search tool?** Give the tool a clear description and say in the system message that it must search before answering. Then check the logs to see that it did. **Can an agent use more than one knowledge source?** Yes. You can connect several tools, such as a handbook index and a pricing sheet, and describe each one so the agent picks the right one. [All AI glossary terms, A to Z](https://learn.ynteractive.com/content/glossary) · [Free prompt templates](https://learn.ynteractive.com/content/prompts)