Player not loading? Watch on YouTube
Mayank Agarwal introduces Microsoft AutoGen through the concepts behind agents and the framework's main components. This opening lesson explains the course roadmap and prerequisites before later lessons build and configure agents. He mentions version 0.7 and uses some diagrams from earlier documentation.
A scheduling example makes the agent idea concrete: check a calendar, create a meeting, invite someone and ask for missing information or help with a conflict. Agarwal then explains how a single agent handles a defined goal and how a team distributes work. Each team member can use its own model and tools. His Claude and GPT comments are examples of assigning tasks, not benchmark results.
The AutoGen section covers agents conversing with humans, configurable interaction patterns and different agent capabilities. Agarwal also identifies the framework's Core, AgentChat, Extensions and Studio layers. Other frameworks receive brief mentions.
For learners following the series, the stated requirements include Python 3.10 or later, VS Code, command-line familiarity and a basic understanding of LLM calls and tokens. The course needs model API keys; Agarwal proposes OpenRouter as an option with limited free access. Installation and environment configuration come in the next lesson.
