Basic RAG retrieves documents and generates answers. Agentic RAG goes further — the AI decides WHAT to retrieve, WHEN to use tools, and HOW to break complex questions into steps.
Agent Capabilities
- Tool calling: Agents can query databases, call APIs, run code, search the web
- Multi-step reasoning: Break complex questions into sub-tasks
- Memory: Remember context across conversations
- Multi-agent: Specialized agents collaborating on complex tasks
Learn AI Agent Engineering
Module 8 covers agent architectures, LangGraph, tool calling, and multi-agent orchestration.