Do I need an account to read them?

No. Every lesson can be read whole with no account. Running a command in the lab needs a free account.

Do they use up credit?

A command that calls the lab’s model uses some of the account’s credit, and a free account has enough each month to follow most lessons. Every command also works with no model, so a lesson can always be finished.

Which model do they use?

The lessons never name one. The platform chooses the model its labs use, and no lesson depends on what it answers.

Have people checked them?

The modules were written by AI agents and their commands were checked against stand-ins for the model, but they have not been run against the real model by a person, and a lecturer has not yet read them. Read a lesson before you set it, as you would any lesson you did not write.

AI lessons: models, agents and AI in production

The AI lessons are five series of undergraduate-level modules, 33 in all, each written separately in English, French, Spanish, Japanese and German. They have a card of their own on the Lessons page, beside Basics, and every lesson can be read whole with no account.

The five series

Series A, AI for software engineers, is for a student or new graduate who can program and use Git: coding with an AI assistant, working with coding agents (writing a task an agent can finish, short context files, tests written first as guardrails, reviewing what it changed, and deciding what not to hand over), calling a model from code, structured output, tool calling, building a chatbot and answering questions from documents.

Series B, AI in production, is for experienced developers: building agents, evals and tests for model-backed features, cost, tracing and reliability, securing the application, architecture and delivery, and a case study of shipping a feature.

Series C, advanced agent systems, is for engineers who have finished series B or have the same experience: harness engineering, several agents working together, evaluating and securing agents, retrieval systems, and governing and changing an AI system.

Series D, applied AI, is for anyone who can program: how a language model works (tokens, attention and a tiny model trained in NumPy, with no GPU), prompting and context craft, images and sound as input, running models yourself, adapting a model, writing an MCP server, designing AI products for people, and text to SQL (letting a model turn a question into a query, then checking that query by reading it, running it read only on data with a known answer and scoring it before anyone trusts it). Every lesson in it has a lab command, a diagram, a matching or crossword activity, a quiz and a checked task.

Series E, running and trusting AI, is for engineers who have finished series B: evaluating AI systems in depth, privacy and personal data in AI apps, operating AI features, change and incident management, and keeping agents safe. It uses invented data throughout and is plain-words teaching, not legal advice.

How a lesson works

Each module is worked in the AI agent lab, except how language models work, which uses the JupyterLab data lab and calls no model. A command in a lesson runs in the student’s own lab, and where it calls a model it goes through the lab’s own connection, so no key is typed anywhere. Every command also works with no model: where the model cannot answer, the script prints a labelled recorded sample and carries on, so a lesson never stops on a refusal.

Assignments are Python programs with hidden tests over fixed data. The public pages show the commands and never run them, and the teacher guide and the tests are never public.

The AI glossary

The AI glossary, linked from the AI lessons page, gives short, plain definitions of about fifty terms a programmer meets, from token and context window to RAG, MCP, evals and prompt injection, and ten side-by-side comparisons such as RAG vs fine-tuning and local model vs API. Each term and comparison has a page of its own in every language, with a one-sentence answer first, a short explanation or table, and links to the lessons that teach it. It is free to read with no account, and names no commercial model.

Using them in a college

Find the AI modules in the Library, or in the Undergraduate computing modules folder in your language. A college copies a whole module into one of its modules with Copy all its lessons, then sets the lessons and assignments as it would any others. Open the teacher guide first: it has the weekly plan.

What it costs, and how to try it

A school or college has its first 365 days free, for up to 50 students and with no card, and everything on this page is part of it. After that the price is per student a year, with staff free and labs, virtual machines and AI included on a fair-use basis. The pricing page has the figure in your currency and a calculator.

Questions people ask

Do I need an account to read them?: No. Every lesson can be read whole with no account. Running a command in the lab needs a free account.
Do they use up credit?: A command that calls the lab’s model uses some of the account’s credit, and a free account has enough each month to follow most lessons. Every command also works with no model, so a lesson can always be finished.
Which model do they use?: The lessons never name one. The platform chooses the model its labs use, and no lesson depends on what it answers.
Have people checked them?: The modules were written by AI agents and their commands were checked against stand-ins for the model, but they have not been run against the real model by a person, and a lecturer has not yet read them. Read a lesson before you set it, as you would any lesson you did not write.