Free Course · 5 lessons · ~3 hours
Build Your First AI Agent
A hands-on introduction to how modern AI agents actually work. By the end you will have built, tested, and broken your own agent — and you will know what to do next if you want to go deeper.
Who this is for
You should take this course if you have heard about AI agents, you are not quite sure what they really are, and you would like to find out by building one yourself. Software engineers and non-developers are both welcome. The exercises are done in your browser using Claude or ChatGPT — there is no installation, no Python, no API keys.
If you have already built agents with LangChain, LangGraph, or a similar framework, you will move quickly. The early lessons will still be a useful mental model; the later lessons cover failure modes that experienced builders often miss.
What you will learn
What is an AI agent?
The actual definition. How an agent differs from a chatbot. The minimum requirements for something to count as an agent.
The agent loop
Think, act, observe — the loop every modern agent runs. You will walk through a real run, step by step, and see where things succeed and where they fail.
Giving an agent tools
Tools are how an agent acts on the world. You will give your agent a calculator, a search, and a pretend email-sender — and watch it learn to use them.
Multi-step planning
Breaking a goal into sub-goals. You will get an agent to plan a small project from scratch and execute it across multiple turns.
When agents go wrong
The most common ways agents fail in the real world — and the patterns that keep them safe. By the end of this lesson, you will have written a one-paragraph governance plan for your own agent.
What you will need
- About three hours of total time. Each lesson is 20 to 40 minutes.
- A browser, and an account with either Claude.ai or ChatGPT. The free tier of either is enough for every exercise.
- That is it. No code, no installs, no API keys.
How this fits with the full programme
This free course is a taster, not a substitute for the Integrated AI Program. It introduces the concepts you would meet in the first weeks of C6 (Generative AI and Foundation Models) and A2 (Large Language Models — Training, Fine-tuning, RAG). At the end of the course, we will tell you exactly which Tier 2 and Tier 3 courses take this material deeper.
You can finish this course without ever applying to the programme. We hope it is genuinely useful regardless of what you do next.
Other free courses
Free course · 5 lessons
Build a Market Research Bot
A hands-on, education-only course on using AI to research companies and markets. Browser-only. Not for trading.
Take this courseFree course · 5 lessons
AI Security Foundations
What every security professional needs to know about AI systems. OWASP, MITRE ATLAS, defence in depth, EU AI Act.
Take this courseFree course · 5 lessons
AI for Sport Analysts
Performance, tactical, scouting, and broadcast analytics. A sport-agnostic course — you bring the sport.
Take this courseFree course · 5 lessons
The EU AI Act for Non-Lawyers
What the EU AI Act actually requires. Risk tiers, provider duties, deployer duties, GPAI rules, and a 30-day checklist.
Take this course