New course on agents! Enroll in AI Agentic Design Patterns with AutoGen

And taught by Chi Wang and Ting Yunwu, who are both creators of Orthogen. I've really enjoyed using Autogen myself, and there have been occasions when it solved the problem in a totally different way. Ch is a principal researcher at Microsoft and Ting Yun is assistant professor at Penn State.

For example, if you want to write a financial article or blog post and the task requires collecting and analyzing data on the performance of different companies, shares and and synthesizing conclusions, this might need days of research and writing code to produce a quality article. This is where a multi agent framework can be incredibly useful. The Autogen agents in this case would assume different roles and Personas, planning, writing and executing code than writing and critiquing the final report. Think of it as breaking a complex task into smaller manageable parts. Writing the report alone can be time consuming, but imagine dividing the task among friends where each friend tackles one piece of task. Autogen agents work similarly, assigning each part of the task to a specialized agent such as a planner, researcher, coder and writer. And these agents can also ask for your feedback at key moments to let you alter the trajectory of their work when needed. So instead of doing all the work yourself, you are the manager of this team of agents and guide them to get the work done.

You might also have heard Andrew describe agentic design patterns like reflection to use planning and multi agent collaboration. Autogen makes all of these design patterns easy to implement and we will go through each of these patterns in different lessons of this course. The example used in this course span work related ones like carrying out research and also some fun ones like store.

In JSON format which you then see how to apply in a customer onboarding application where your agents will use the information they gathered about the new customer to customize the content shown to them. You also build a group of agents that can reflect on a blog post writing process. This involves a sequence of chats between different reviewers who then report back to a higher level critic agent. This type of chat structure is also called a nested chat and it functions like the inner monologue of that critic agent whose job it is to review and reflect on the blog post and provide suggestions on how to improve the writing.

As a fun example, you will apply a similar nested chat structure to facilitate a conversational chess match by using a tool that defines and understands the legal moves in a chess game. And you will also learn how to add coding capabilities to your agents while making sure that human feedback is also in the loop. You will learn about the best practices for how to structure multi agent group chats and the planning required to build and customize multi agent systems.

Agentic design patterns have triggered a huge interest in the developers community. It is changing every industry from accounting, airlines, blockchain, biotech to retailer, social media, software, supply chain, telecom, from big companies to startups. Developers really want to implement these patents easily, which is what Autogen does. Autogen is used by companies, organizations and universities from all over the world. Some even use Autogen as a backbone to build their own agent platform. Others use it for software development, consulting, research, data processing, gaming and simulation, and so on. I hope the course can help you join this revolution.

I think you might be surprised at not just how productive working with Autogen can be, but also how fun it is. Autogen is a framework worth learning about, so I hope you sign up for and enjoy the course.