Agentic Workflows

In this chapter, we will explore agentic workflows. They are essential for building applications that exhibit dynamic behavior, responding intelligently to changes in the application state.

To grasp the concepts of agentic workflows, we will explore workflow state management, learning how applications transition through different states within a graph. You will then investigate controlled generation techniques, enabling you to specify the desired format for responses from large language models (LLMs).

Then, you will implement agentic RAG, a powerful approach that leverages agents to enhance traditional retrieval augmented generation (RAG) pipelines.

Finally, you will learn how to build agents capable of translating natural language into SQL queries, allowing you to interact with databases using conversational language.