What you'll learn

Learn how to build an agent that can reason over your documents and answer complex questions.
Build a router agent that can help you with Q&A and summarization tasks, and extend it to handle passing arguments to this agent.
Design a research agent that handles multi-documents and learn about different ways to debug and control this agent.

Course recap

PRO

This course teaches how to build agentic RAG systems using LlamaIndex, progressing from simple routers to full multi-document agents. Students start with a router that dynamically selects between vector search and summarization query engines, then advance to tool-calling agents, and finally build a multi-document research agent.

Concept map

Key concepts

Router query engineDynamically selects between query engines (vector search vs summarization) based on the query
Retrieval-Augmented GenerationEnhanced with agentic capabilities for dynamic query understanding
Query toolsQuery engines wrapped with metadata describing what types of questions they can answer
LLM selectorPrompts an LLM to choose which query engine to route to, outputting JSON
Pydantic selectorUses function calling APIs to produce structured selection objects
Summary indexReturns all nodes for comprehensive summarization
Vector indexReturns most similar nodes by embedding similarity for specific questions

Lesson highlights

1.Router Query Engine — Dynamic routing between vector search and summarization over a single document
2.Tool Calling — Agents that select and call tools based on query analysis
3.Building an Agent Reasoning Loop — ReAct-style reasoning with tool use
4.Multi-Document Agent — Research agent spanning multiple papers with per-document tools
5.Advanced Patterns — Combining agents with memory and planning
6.Conclusion

About this course

Join our new short course and learn from Jerry Liu, co-founder and CEO at LlamaIndex to start using agentic RAG, a framework designed to build research agents skilled in tool use, reasoning, and decision-making with your data.

In this course:

Build the simplest form of agentic RAG – a router. Given a query, the router will pick one of two query engines, Q&A or summarization, to execute a query over a single document.
Add tool calling to your router agent where you will use an LLM to not only pick a function to execute but also infer an argument to pass to the function.
Build a research assistant agent. Instead of tool calling in a single-shot setting, an agent is able to reason over tools in multiple steps.
Build a multi-document agent where you will learn how to extend the research agent to handle multiple documents.

Unlike the standard RAG pipeline—suitable for simple queries across a few documents—this intelligent approach adapts based on initial findings to enhance further data retrieval. You’ll learn to develop an autonomous research agent, enhancing your ability to engage with and analyze your data comprehensively.

You’ll practice building agents capable of intelligently navigating, summarizing, and comparing information across multiple research papers from arXiv. Additionally, you’ll learn how to debug these agents, ensuring you can guide their actions effectively.

Explore one of the most rapidly advancing applications of agentic AI!

Who should join?

Anyone who has basic Python knowledge and wants to learn how to quickly build agents that can reason over their own documents.

Building Agentic RAG with Llamaindex

What you'll learn

Course recap

Concept map

Key concepts

Lesson highlights

About this course

Who should join?

Course Outline

Elevate your learning experience with Pro

Instructor

Jerry Liu

Building Agentic RAG with Llamaindex

Want to learn more about Generative AI?