liveBook · Manning

Agentic RAG

Agentic RAG is a retrieval-augmented generation approach in which an agent dynamically decides how and when to retrieve information, often iterating or using tools before generating a response.

How 9 Manning books explain it

Retrieval Augmented Generation, The Foundational Ideas

Agentic RAG is a retrieval-augmented generation architecture in which runtime decisions control retrieval, reasoning, or tool actions based on system state and signals.

How it is used
Addresses questions whose evidence needs change during answering, especially when a missed document or incomplete answer is costly.
How deep it goes
Defines runtime control, distinguishes it from fixed routing, and implements a capped retrieval–evaluation loop; leaves evaluation helpers outside the shown listing.
Examples
Uses a multi-source compliance question, a Python retrieval loop, and an offline fintech comparison to expose benefits and costs.
What you can do after
Build it from scratch as a capped retrieval loop, then evaluate whether its accuracy gain justifies extra latency against a fixed-pipeline baseline.

Read how Retrieval Augmented Generation, The Foundational Ideas explains agentic RAG →

AI Agents in Action, Second Edition

How it is used
Uses agent-controlled retrieval to answer questions from external knowledge and support larger workflows.
How deep it goes
Develops vector and hybrid agents through indexing, tool instructions, grounding, guardrails, and evaluation; omits hybrid tool implementations.
Examples
Uses a movie-script search agent, benchmark questions, and a customer-support workflow.
What you can do after
Apply it via the OpenAI Agents SDK and Chroma DB, with retrieval tools, grounding instructions, and failure handling.

Read how AI Agents in Action, Second Edition explains agentic RAG →

Domain-Specific Small Language Models

A retrieval-augmented generation setup that uses an AI agent to plan, select, and combine retrieval actions across one or more sources before generation.

How it is used
Uses an agent to plan and adapt retrieval across sources before generating an answer.
How deep it goes
Works through PDF preparation and BM25, semantic, hybrid, and web-search tools, but does not detail agent planning internals.
Examples
Uses a chunked PDF and smart-home queries to demonstrate local retrieval, web search, and multiple-tool selection.
What you can do after
Apply it via smolagents CodeAgent with local and web-search tools, including a weighted hybrid retriever.

Read how Domain-Specific Small Language Models explains agentic RAG →

LLM Evaluation and Alignment, The Foundational Ideas

An agentic RAG system is a retrieval-augmented generation system that autonomously formulates retrieval queries, evaluates results, and decides whether to continue retrieving.

How it is used
Deploys retrieval as an iterative reasoning tool for resolving information gaps before generating an answer.
How deep it goes
Explains the reason-act-observe loop and trace-level evaluation, while stopping short of implementation details for query formulation.
Examples
Uses a financial research assistant and a multi-step retrieval trace as concrete examples.
What you can do after
Diagnose retrieval, generation, and reasoning failures by evaluating queries, retrieved context, intermediate traces, and final answers.

Read how LLM Evaluation and Alignment, The Foundational Ideas explains agentic RAG →

Context Engineering

Agentic retrieval-augmented generation uses an AI agent to plan, perform, and evaluate retrieval while producing an answer.

How it is used
Uses an agent to orchestrate external knowledge retrieval and context construction for grounded answers.
How deep it goes
Explains iterative retrieval and source selection, then demonstrates a minimal single-tool agent rather than implementing the full multi-source architecture.
Examples
Illustrates a shopping assistant comparing phone specifications and a private document search for a book’s author.
What you can do after
Apply it via LangChain by wrapping a retriever as a tool and configuring a tool-calling agent.

Read how Context Engineering explains agentic RAG →

Designing AI Systems

Agentic RAG is a retrieval-augmented generation approach in which a model decides when and how to retrieve information and can iterate on searches.

How it is used
Grounds organizational answers by giving the model iterative access to searchable company knowledge.
How deep it goes
Explains tool registration, result scoring, query reformulation, and semantic ingestion, but stops short of implementing model reasoning.
Examples
Demonstrates Q2 planning searches, engineering commitments, relevance scores, alternate indexes, and an organizational knowledge index.
What you can do after
Apply it via a registered knowledge-search tool backed by indexed documents and model-controlled iterative retrieval.

Read how Designing AI Systems explains agentic RAG →

Essential GraphRAG

Agentic RAG is a retrieval-augmented generation pattern in which agents and routing mechanisms select and coordinate retrieval actions to answer a question.

How it is used
Routes questions across multiple data sources and validates retrieved answers when generic retrieval is insufficient.
How deep it goes
Defines routing, retrieval, and answer criticism, while stopping short of advanced multi-step task planning.
Examples
Uses generic vector retrieval, specialized text2cypher, and parameterized database queries as alternative retriever agents.
What you can do after
Build retrieval workflows that select suitable sources and assess answer quality before returning results.

Read how Essential GraphRAG explains agentic RAG →

Building Reliable AI Systems

A retrieval-augmented generation approach in which an AI agent controls the retrieval process, including deciding when multiple retrievals are needed.

How it is used
Combines document retrieval with agent-directed searches, external tools, and real-world tasks.
How deep it goes
Explains multi-source retrieval decisions and practical tool integration through a travel-assistant implementation.
Examples
Uses policy-and-review comparisons and travel planning with a weather service to illustrate retrieval needs.
What you can do after
Build a travel assistant that combines document retrieval with real-time services using LangChain.

Read how Building Reliable AI Systems explains agentic RAG →

Designing AI Agents

Retrieval-augmented generation in which an agent can make decisions about whether and how to retrieve information.

How it is used
Lets a model decide whether and how to retrieve evidence across sources.
How deep it goes
Implements query rewriting, hybrid search, fusion, and context assembly within a broader governed pipeline.
Examples
Uses query rewriting, hybrid search, result fusion, and context assembly as an implementation example.
What you can do after
Apply it by letting a model choose whether to retrieve, revise queries, and route across sources.

Read how Designing AI Agents explains agentic RAG →

Also covered in

What to know first

Related concepts

Questions about agentic RAG

What is agentic RAG?

Agentic RAG is a retrieval-augmented generation approach in which an agent dynamically decides how and when to retrieve information, often iterating or using tools before generating a response.

Which Manning books explain agentic RAG?

Retrieval Augmented Generation, The Foundational Ideas, AI Agents in Action, Second Edition, Domain-Specific Small Language Models, LLM Evaluation and Alignment, The Foundational Ideas, Context Engineering, Designing AI Systems, Essential GraphRAG, Building Reliable AI Systems and 1 more.

What should I know before learning agentic RAG?

Manning books that teach agentic RAG usually introduce retrieval-augmented generation, similarity search, AI agent and accuracy first.

What is agentic RAG a kind of?

Agentic RAG is a kind of retrieval-augmented generation. Retrieval-augmented generation is a technique that retrieves relevant information from external sources and supplies it as context to a generative model when producing an output.

These guides are generated from the text of Manning books by Manning's knowledge graph and link to the sections they draw on. The full books are available on liveBook and at manning.com.