AI Agents & Workflows - The Practical Guide

COURSE UPDATE

This course was updated to reflect the change in what agents are today, more relevant capabilities, and emerging frameworks for building such agents.

The course is now less about code deep dives, more about the core concepts (and how to apply them when using or building agents).

---

"AI Agents" is a buzzword!

That's a pity — because real AI agents (and AI-powered workflows) can unlock a huge range of opportunities. From summarizing documents and generating files, to customer support assistants, research helpers, and automations you wire into tools like Slack. Your imagination is the limit.

This course cuts through the hype and shows you what AI agents actually are - how they work behind the scenes, how they differ from AI workflows, and how you can use and build both confidently.

You'll look at agents from two angles: as a user of tools like Claude Cowork / ChatGPT-style agents, and as a builder who wants to plan or create agents visually or in code. The goal is a deep, practical mental model — not empty buzzwords.

With this course, you'll get:

• A clear explanation of what AI agents are (and what they are not)

• The difference between AI workflows and AI agents — and when to choose which

• The core building blocks: models, harnesses, the agent loop, tools, instructions, context/sessions, skills, memory, sandboxes, and humans-in-the-loop

• How general agents differ from task-specific agents, and where agents can run (your machine vs the cloud)

• Concrete examples you can follow: visual builds with n8n, code examples in Python, plus framework-based builds

• Guidance that transfers beyond one stack — so you can apply the ideas with the models and tools you already use

You'll learn:

• What an agent harness is and how the agent loop really works

• How LLMs use tools (and what models can and cannot do on their own)

• How to steer agent behavior with system instructions, AGENTS .md / CLAUDE .md, and agent skills (progressive disclosure)

• How to manage context, sessions, compaction, and longer-term memory concepts

• How to build an AI workflow visually (e.g. a document summarizer in n8n) and the same idea in code

• How to build AI agents visually (tools like summarize text & create PDF) and step-by-step in code

• How “primitive” tool calling works vs provider-native tool calling

• How more general agents support files, commands, project instructions, and skills

• How to constrain agents with sandboxes and human approval (e.g. customer-support style refunds)

• How frameworks speed this up — including a deep dive into Eve, a look at CrewAI, and an overview of other options (e.g. Vercel AI SDK, LangGraph, Pydantic AI)

• How a task-specific agent can be exposed beyond your laptop (including a Slack demo)

This is not primarily a programming course! You don’t need to be a developer to follow along: the code and visual projects are there so you can see how agents work. If you do build software, the same foundations apply whether you use Python, TypeScript, n8n, or a framework.

By the end, you won’t just recognize the buzzword “AI agent” - you’ll understand the moving parts, know when a simple workflow is the better choice, and be ready to work with agents productively or design your own.