Sample course · Meme style · 4 lessons · about 22 min
How AI agents work
What an AI agent actually does, how it uses tools, why it goes wrong, and how to use one safely.
Start lesson 1 Free to read. No account needed.
An AI agent is a language model running in a loop: it asks for tools, reads the results and decides the next step until the task is done.
More apps now offer agents that search, edit files or act in your accounts. Knowing how they work tells you what to hand over, what to approve and what to check.
Who it's for
Anyone who uses AI chatbots and wants to know what an agent really does before trusting one with a task.
Not for: People who want to build or code their own agents, or compare specific products.
Before you start
Nothing. We start from the beginning.
You'll be able to
- Explain what an AI agent is and how it differs from a chatbot that just replies
- Describe how an agent calls tools and when it stops
- Spot common agent failures and set it up to be safer
What this course covers
- What an agent is, and the fuzzy line with chatbots
- The loop: think, act, look
- Tools and tool calling
- When an agent stops
- Compounding errors and stuck loops
- Made-up tool results and context limits
- Planning and memory
- Least privilege and asking before irreversible steps
- Prompt injection
- Checking the work
- Multi-agent systems · not here: Several agents passing work between them builds on the same loop; not needed to understand or use one
- Building an agent in code · not here: This course is for using and judging agents, not programming them
Lessons
- 1Getting started: what an AI agent actually isSee what an agent is, where the model and tools fit, and how it differs from a chatbot that just replies. 5 min
- 2Tool calls and the loop: one more step, I promiseLearn how an agent asks for a tool, what happens to the result, and what makes it stop. 5 min
- 3Off the rails: why agents go wrongSee how small errors add up over many steps, why agents get stuck or invent results, and what planning and memory do. 6 min
- 4Trust issues: using agents safelyLearn least privilege, when an agent should ask first, what prompt injection is, and how to check the work. 6 min
Same app, other styles
- How AI chatbots actually work · Meme
- How a bill becomes law · Story
- Compound interest, with real numbers · Classic
Want one on your own topic or notes? Make a course free