Learn

What to learn before you start building with AI

Learn what each part of an app is for, not how to write it: the frontend and backend, APIs, environment variables, a database, login, webhooks, Git, deployment and DNS. You don't need a programming language. You need enough of each idea to follow what your AI is doing and notice when its answer skips a step.

The table puts them in roughly the order a first project runs into them. Read each guide when you get there; reading all nine before you start is fine, but not necessary.

What you can safely skip

You don't need to memorise code syntax, compare programming languages or learn a framework before you start. Your AI coding agent handles that part, and that part keeps changing as the tools change. What doesn't change is the shape underneath: something the visitor sees, something that runs where they can't reach it, somewhere the information is stored, and a way to get it all online. That shape is what this hub teaches, because it stays true whichever tool you use.

Learning these ideas has a limit worth being honest about. It won't let you review your app's code line by line, and it doesn't replace testing. What it does is let you ask the right question ("where is that key stored?", "what happens if the payment message arrives twice?") and recognise when an answer sounds sure of itself but leaves something out.

Choosing a tool comes before the concepts

If you haven't started yet, pick the kind of tool before anything else. How to choose a vibe coding tool explains the three kinds (a chat artifact built in a panel beside your chat, an AI website builder, and an AI coding agent that changes the actual files of a project) and the four questions that decide between them. What Is Vibe Coding? covers what the practice can and can't do for a non-coder, and What Is an AI Agent? explains why many AI build tools already work as agents, taking several steps on their own.

Claims worth checking before you believe them

A lot of what beginners hear about building with AI comes from posts written to go viral: build a business in a weekend, one prompt replaces a developer, a secret trick gets you to the top of Google. Some of it is partly true, and most of it leaves out the part that takes the time. Do SEO hacks actually work? checks the search-ranking claims one by one. The honest version is usually the same: the result is possible, but it takes the steps the post skipped.

Getting found on Google: the fundamentals

Schema markup and breadcrumbs are how a search engine learns what a page is and where it sits on your site. Neither is a hack, and neither replaces useful content, but getting them wrong is a common, avoidable mistake on AI-built sites. Schema Markup Types Explained and Breadcrumb Navigation vs. Schema Markup explain the ideas; the two prompt pages give you copy-paste prompts that get your AI to write the markup correctly and then check it.

A prompt for any term your AI uses

When your AI coding agent uses a word you don't know, don't nod along. Paste this and fill in the word:

I'm not a coder. You just used the term [paste the
word, e.g. "environment variable" or "webhook"].
Before we go on:

1. Explain what it means in two or three plain
   sentences, with no other jargon.
2. Tell me what it does in my project specifically:
   which file or setting it lives in, and what would
   break if it were missing or wrong.
3. Tell me if I need to do anything myself (add a
   value, change a setting outside this chat), or if
   you can do all of it.

If you're not sure how my project is set up, ask me
instead of guessing.

What to check in the AI's answer

• It points to something real in your project. A named file, setting or screen, not "somewhere in your configuration".
• It says what breaks without it. If it can't say, ask again; that's the part you need when something goes wrong later.
• It's clear about who does what. Anything you have to do outside the chat (add a value on your host, change a setting with a service) should be spelled out, because your AI can't do those for you.
• It matches the guide. If the answer disagrees with the guide in this hub on the same idea, ask it why before accepting either.

Where to go next

Once you know the words, Prompting covers how to direct your AI well: writing clear instructions, describing a bug, and getting out of a fix-break-fix loop. Building applies these ideas to your own project, from adding login to deciding when you need a database.