Welcome to AI Agentic Design Patterns with AutoGen, built in partnership with Microsoft. I'm joined by the instructors Chi Wang, who is principal researcher at Microsoft, and Qingyun Wu, who's an assistant professor at Penn State University, both of whom are co-creators of AutoGen. Thanks, Andrew. I'm excited to be here. Thank you. Great to be here with you, Andrew. This course is an introduction to AutoGen, a multi-agent conversational framework that enables you to quickly create multiple agents with different roles, persona tasks, and capabilities to implement complex AI applications using different AI agentic design patterns. Let's say you're interested in analyzing financial data. The task may require writing code to collect and analyze share prices. Then synthesizing your findings into a report. This might take a person days of research and coding and writing. A multi-agent system can streamline this process by enabling you to create and hire agents that work for you as a researcher, data collector, co-writer, and executor. Your agents can also iteratively review, critique, and improve the results until it meets your standard. This is just one example of the many practical applications of a multi-agent framework. And Chi and Qingyun will guide you through six lessons, each featuring its own unique design process and use case. In this course, you'll learn about how to train core components by exploring the building agent conversable agent. You'll create and customize a lively conversation between two standup comedian agents while exploring their interactive capabilities. Then you will learn about a multi-agent interaction pattern called Sequential chats. This allows you to build conversational agents that work step by step to carry out a list of tasks in a sequence. We'll illustrate this with a customer onboarding application. You will also explore the agent reflection framework in two practical scenarios. The first one will use multiple agents to produce a well-written blogpost, the second as tools to agents to create a conversational chess game. In both of these examples, you will learn another interaction pattern called a nested chat, which means when the agent is given a task, it calls a bunch of other agents and iterates with them for a while before it returns the result. If we think of you, the developer, as a manager of a handful of agents, then nested chat corresponds to letting an agent that you manage also in turn, manage their own agents. You also learn about a very powerful capability tool use. In which you can provide a user-defined function. For example, a function to check of certain chess moves are legal and give that to the agent to use. And for applications where you might not have a predefined function or predefined piece of code to provide, you'll also learn about coding and code execution, which means asking the agent to write the code it needs, which after checking for correctness, it could then execute that code it did wrote. Maybe in a sandbox environment to computer result needed for a task. You'll see both tool use as well as code writing and execution illustrated in the financial analysis example. And finally you learn best practices for building custom multi-agent group chats. We'll illustrate this with examples of generating a detailed report. This complex task requires planning. In other words, the agent has to decide on a sequence of actions to take, such as guiding data, analyzing, writing, and revising. You see how we can add a planning agent into the group chat and control how the work flows from one agent to another agent, so as to execute the needed individual tasks in the right sequence. Many people have worked to create this course. I'd like to thank Eric Zhu from Microsoft. As well as Diala Ezzeddine from DeepLearning.AI. In this first lesson, you will start with building your first AutoGen agent. Program a basic two-agent conversation, and enjoy a standup comedy show. That sounds great. Let's go on to the next video and get started.