Learn AI architecture

Six interactive modules on how AI systems are put together, written for people who make decisions about them. No code required, no hype supplied.

This is a free course on how AI systems are put together: what a model is, how retrieval grounds one in your documents, why data quality decides outcomes, what changes when AI acts rather than answers, how to govern the whole, and how to buy any of it well. Six short modules, each with something to click on, all in plain English, none of it needing a technical background.

I wrote it because I kept needing it. My work is helping organisations through change, AI adoption included, and the same requests kept arriving: explain embeddings before Thursday's supplier meeting, explain agents before the board's risk review. These modules are those explanations, written down properly, with the diagrams I would otherwise draw on a whiteboard.

Some ground rules. No sign-up and nothing gated. No product names in the teaching, because they age in months; anything fast-moving sits in dated boxes I review quarterly. And every module says when not to use the pattern it teaches, because that is usually the most useful part.

  1. Module 1

    Foundations

    Models, tokens, context windows, and the journey of a prompt: what is happening inside the machine you are paying for.

    Start the module

  2. Module 2

    Retrieval-Augmented Generation

    How AI systems answer from your documents instead of their memory, with a live retrieval demo running in your browser.

    Start the module

  3. Module 3

    Data foundations

    Lakes, warehouses, and pipelines: why AI projects stand or fall on data decisions made years earlier.

    Start the module

  4. Module 4

    Agents and orchestration

    What changes when AI systems act rather than answer: tool use, multi-step workflows, and the guardrails they demand.

    Start the module

  5. Module 5

    Governance

    Where risk lives in everything the first four modules built, and who is accountable for it. The module the others are building towards.

    Start the module

  6. Module 6

    Buying AI

    The argument applied to procurement: self-host or buy access, what suppliers may do with your data, true costs, and the ten questions to ask.

    Start the module

Course references: the glossary, every term the modules teach in plain English, the anatomy of an AI system, the whole architecture with toggleable layers, and who you need, the roles an AI programme takes and which of them write code.