1 The rise of AI agents

By themselves, LLM-based apps, like simple chatbots, can generate responses and answer questions. Next, we want them to make plans and carry them out. We want them to book a flight, not just list available flights, or update a project tracker, not just list the changes that need to be made. To equip AI apps with active abilities, we add AI agents to the system. The agent is not a new concept in machine learning and artificial intelligence, and the term can be ambiguous. But when we talk about intelligent or AI agents, we generally mean software that perceives its environment, decides what to do, and takes action to achieve a goal, using the resources provided by an LLM.

1.1 Defining agents and agentic thinking

1.1.1 Understanding agent, assistant, and LLM patterns

1.1.2 Thinking like agents: Sense-plan-act-learn

1.1.3 Agents act with tools

1.2 Introducing the Model Context Protocol

1.3 Understanding the five functional layers of an agent

1.3.1 The agent persona

1.3.2 Agent tools and actions

1.3.3 Agent reasoning and planning

1.3.4 Agent knowledge and memory

1.3.5 Agent evaluation and feedback

1.4 Advancing to multi-agent systems

1.4.1 The agent flow assembly line

1.4.2 Agent orchestrations (hub-and-spoke)

1.4.3 Agent collaboration (teams of agents)

1.5 Next steps

Summary