Summary
This chapter explored the core concept of “agent” in generative AI applications. We delved into how agents utilize tools, strategically planning their use to achieve goals. We explored the components of an agent: LLMs, prompt instructions, memory, intermediate steps, and tools.
LangChain’s agent implementation was demonstrated, illustrating the building of agents using LangGraph and tool integration.
Finally, we introduced Vertex AI Agent Builder, detailing its features and how to create agents using it. The chapter concluded with a brief overview of extensions in Vertex AI Agent Builder, showcasing its ready-to-use tools, such as code interpreters and data stores. In the next chapter, we will explore how an agentic workflow works and give some examples of them, such as natural language to SQL and agentic retrieval-augmented generation.
