Free course
Loop & Agentic Engineering
Move beyond single prompts into self-correcting loops and multi-step agents: critique-revise cycles, the ReAct pattern, tool use, and knowing when to stop a loop.
What you'll be able to do
You will be able to design AI systems that run more than one step without falling apart — chaining calls, feeding output back as input, deciding when to stop, and handling the case where an intermediate step goes wrong.
This is the gap between "AI helps me write" and "AI does a task end to end". You will finish with a working multi-step loop you built yourself, and an understanding of why per-step reliability compounds: a system that is 95% right at each step is under 60% right across ten.
Who it's for
People who have used AI for single tasks and now want it to handle a process. Some technical comfort helps — you will be reasoning about control flow and state — but this is not a programming course and the loops can be built in an automation tool rather than in code.
Who it isn't for
If you are not yet writing prompts that reliably produce structured output, start with Prompt Engineering Fundamentals. Chaining unreliable steps together multiplies the unreliability, and this course will be frustrating rather than useful.
What you'll need
Prompt Engineering Fundamentals, or equivalent experience writing structured prompts. Access to an AI tool that lets you run several calls in sequence — an automation platform, a notebook, or the API.
How it's taught
Built around failure modes rather than architectures. Each lesson introduces a way multi-step systems break — drift, runaway loops, silent errors, context bloat — and then the technique that addresses it. Understanding why something is necessary before learning it makes it far more likely you will apply it correctly.
What you'll cover
4 modules, 6 lessons. Every lesson includes copy-ready prompts tested against the models they're written for.
Module 1 — What Is a Prompt Loop
- Critique-and-Revise Loops
Module 2 — Building Agent Loops
- The ReAct Pattern (Reason + Act)
- Giving Loops Memory
Module 3 — Advanced: Multi-Agent Patterns
- Planner–Worker–Reviewer Teams
- Loop Budgets and Failure Handling
Final Project — Ship a Self-Correcting Loop
- Project: Draft–Critique–Revise Content Machine
Finish the lessons and pass the end-of-course quiz to earn a shareable certificate.
Ready to start?
Create a free account to unlock the lessons, track your progress and save prompts as you go.
Start free