How to Use AI Agents for Total Beginners: A Crash Course w/ Agent Builder James McAulay

Building a useful AI agent requires more than connecting an AI model to a few tools. The agent needs verified context, clearly defined permissions, repeatable instructions, and a reliable trigger for beginning its work.

In this crash course, agent builder James McAulay explains how he uses Claude Cowork and Claude Code to build practical AI agents. James breaks down how context files, MCP connectors, reusable skills, safety controls, testing, and scheduled tasks work together.

You’ll see how these components can turn an ordinary AI conversation into a workflow that returns completed, reviewable work. The discussion also covers failed demonstrations, local and cloud scheduling, permission risks, model costs, and the decisions that should remain under human control.

What you'll learn: 

  • Choose a narrow, repeatable workflow with a clear finished result.
  • Organize and verify the context an AI agent needs to perform useful work.
  • Connect external tools through MCP while limiting unnecessary permissions.
  • Turn a proven workflow into a reusable skill and test it with realistic examples.
  • Schedule a reliable agent while preserving human review and approval.
You don’t need to build a fully autonomous agent or reproduce James’s complete technology stack. This crash course focuses on a practical framework you can apply to one recurring task using the AI tools and connected applications available to you.
  • 2 hours of video instruction
  • 1 lesson
Patrick Jones - Course author
Meet YOUR instructor

Grant Harvey

Grant is the voice of The Neuron, the daily newsletter delivering no-hype insights on how AI is changing the world around us, along with co-hosting The Neuron’s weekly podcast.