Building AI Agents with AutoGen

Course details

In this comprehensive, Harshit Tyagi—a data science instructor and mentor—equips you with the skills to develop AI agents using the AutoGen framework. Covering key design patterns such as Reflection, Tool Use, and Multi-Agent Collaboration, this course enables the creation of multi-agent systems capable of handling complex tasks autonomously. Through hands-on projects, you will integrate external tools and refine agent capabilities with human feedback. Ideal for intermediate to advanced users with basic Python experience, this course prepares you to automate complex workflows effectively.

Learning objectives
Build multi-agent systems using the Autogen framework to handle complex tasks autonomously.
Implement and utilize key agentic design patterns such as Reflection, Tool Use, and Multi-Agent Collaboration.
Customize and deploy AI agents capable of interacting and collaborating to solve real-world problems.
Integrate external tools and APIs within agent workflows to enhance functionality and performance.
Leverage human feedback to refine and improve agent capabilities, ensuring robust and accurate outputs.

Concepts

Introduction

Let's build some AI agents with AutoGen

AI Agents with AutoGen

Understanding AI agents
Overview of the AutoGen framework
Configuring LLMs and setting up AutoGen
Building a basic

Developing a Research Assistant

Multi-agent conversation
Research assistant design
Creating agents for the assistant
Creating tasks for the agent

Solving Complex Tasks Using Nested Chats

Nested chats to add multi-level reviewing
Creating reviewer agents
Creating nested chat

Tool Use and Adding Internet Capability

What are tools
Add calculation tool to math tutor agent
Register tools and run the agent
Add web search ability to the research assistant

Building an End-to-End Coding Copilot in Autogen

Code executors in AutoGen
Create agents
Define task to analyze data
Execute user-defined code in AutoGen

Teachable Agents

Introduction to teachable agents
Setting up teachable agents
Teaching and retrieving information

Building a Data Analysis Company with Group Chat

Organizational structure of agents
Build the group chat
Speaker selection config

Conclusion

Next steps with AutoGen