Build to learn: master AI by creating a real website

This module exists to show you a single, powerful truth: AI doesn’t read your mind. It fills in the gaps you leave with its best guess, and its best guess is usually average. The quality of what AI produces is entirely determined by the quality of what you ask for. This isn’t about memorising prompt templates. It’s about understanding why specificity matters, so you can apply the principle to any AI tool, any task, any domain. What you’ll learn to do (not just know) • Recognise the difference between a vague request and a complete specification • Understand why AI defaults to generic output (statistical plausibility) • Identify where ambiguity in your instructions leads to wrong implementations • Use the universal spec format to structure any AI request • Articulate what “done” looks like before you start building

You know what you’re building. You know what “done” looks like. Now you’ll build it — and in the process, learn that the hardest part of working with AI isn’t getting it to produce something. It’s getting it to produce the right thing. This module teaches you two complementary workflows: writing a spec so good you get it right first time (the specification approach), and iterating effectively when you don’t (the iteration approach). The best AI users know when to use each. What you’ll learn to do (not just know) • Translate a project specification into a prompt using the universal spec format • Distinguish between the specification workflow (get it right in one shot) and the iteration workflow (refine through conversation) • Iterate on AI output with targeted, specific feedback — not “make it better” • Identify and fix AI-isms: generic text, placeholder content, and statistical defaults • Recognise when a problem is in your specification vs. a genuine AI limitation

You can now write a detailed specification and iterate effectively. But writing specifications takes time and thought. What if you could get AI to help you write better specifications? This is meta-prompting — using AI to improve your own use of AI. It’s where the real leverage appears, and it’s where most people stop being AI users and start being AI directors. What you’ll learn to do (not just know) • Use meta-prompting to generate structured specifications from conversational descriptions • Apply the 90/10 workflow: AI for structure, human judgment for domain knowledge • Critically evaluate AI-generated specifications instead of passively accepting them • Understand the four layers of AI direction and where you sit on the progression

You’ve built a website, but that was never really the point. The website was a training ground for a set of skills that apply to every AI interaction you’ll ever have. This module makes sure those skills transfer — that you leave this course able to work effectively with AI on anything, not just web projects. It’s also about the reality that AI capabilities change constantly. The skill that matters most isn’t mastering today’s tools; it’s knowing how to stay at the frontier. What you’ll learn to do (not just know) • Apply the universal spec format and iteration skills to a non-website task • Understand the expanding bubble model of AI capabilities and how to stay current • Develop boundary sensing — knowing what AI handles well today vs. where it fails • Design clean handoffs between AI work and human judgment (seam design) • Choose deliberately between the specification and iteration workflows based on task type

You can direct AI effectively. Now make sure the job market knows it. This module translates the skills you've built — specification writing, constraint architecture, critical evaluation, seam design — into the language employers actually use, and teaches you to tell the difference between companies that genuinely understand AI and companies that just hand out ChatGPT subscriptions. You'll also learn to ask the questions most candidates don't know to ask, like whether an organisation uses a specification or iteration workflow as their standard — a single question that reveals more about AI maturity than any job listing ever will. What you'll learn to do (not just know) Translate AI direction skills into CV, LinkedIn, and interview-ready language that maps to recognised competencies like requirements definition, workflow design, and quality assurance Ask interview questions that reveal an employer's real AI maturity — not just which tools they've bought, but how they evaluate output, handle iteration, and invest in skills Identify red and green flags in how organisations talk about AI adoption Build a specific, testable 90-day plan for integrating AI skills into your current role or job search — written with the same acceptance criteria rigour you'd apply to any spec