Building AI Agents that actually work (Full Course)

The podcast explores how to leverage AI agents to automate business departments, moving beyond basic chat models to goal-oriented AI. Remy Gaskell explains the concept of an AI agent, contrasting it with simple question-and-answer chatbots, and details the "agent loop" of observe, think, and act. He emphasizes context engineering, using local files and structured markdown files such as agents.md and memory.md to train AI agents. Gaskell also introduces Model Context Protocol (MCP) as a translator between AI and various tools like Gmail and Notion, enabling agents to perform complex tasks. The discussion covers practical examples, including building skills for specific processes, such as ad analysis and newsletter creation, to automate repetitive tasks and improve productivity.