AI Engineering line, station 2 of 8
Build and Launch a Site with AI
Go from zero accounts to a live website with sign-in and a database, by giving an AI coding agent clear instructions and checking each step yourself.
- 5lessons
- 2.5 hto read and do
- Beginnerlevel
- 5podcast episodes
Tools you'll use in this course
The lessons open these with a template ready. They work for your own job too, no course needed.
About this course
This course is a hands-on walkthrough of the exact process Kolom itself was built with: one person, no coding background, an AI coding agent, and free or cheap services. You make the accounts and install the tools, start a Next.js project by talking to the agent, save every step to GitHub, add a Supabase database with sign-in, and put the site live on Vercel.
You build Mangrove Kitchen, a small recipes-and-offers site for Mangrove Market, the fictional grocery chain used across Kolom. Visitors read recipes built around the week's offers; signed-in members save favourite recipes and choose to get a weekly offers e-mail (the e-mail itself comes in the next course, Run and Grow Your Site). You can build your own idea instead: the steps are the same.
You never write code by hand. You give the agent short, specific instructions, read what it changed, click through dashboards, paste the SQL it wrote into Supabase, and copy settings between services. Every lesson is a numbered walkthrough with safety rules where they matter: two-factor sign-in on every account, secret keys only on the server, row-level security on every table, and what to do when a key leaks.
What you will be able to do
- Set up a safe builder's workbench: VS Code, Git, Node, an AI coding agent, and GitHub, Vercel, Supabase and Google accounts with two-factor sign-in
- Start and grow a Next.js site by giving an AI agent small, checkable instructions and reading what it changed
- Save every working step to GitHub with clear commits, keep secrets out of the repository and undo a bad change
- Create a Supabase project, run the agent's SQL migrations in the SQL editor, turn on row-level security and set up sign-in with the right Site URL and redirect URLs
- Put the site live on Vercel from GitHub, with environment variables, preview links, redeploys and a tested rollback
- Spot the moments that need extra care: a leaked key, a missing access rule, a paid plan you didn't mean to start
Syllabus
5 stops, about 2.5 h in total. Each stop fills in when you complete the lesson.
Your workbench: accounts and tools
Before the AI agent writes a single line, you need a safe workbench: five accounts, four tools on your laptop and a habit of locking every door. This lesson sets it all up in about an hour, the same way Kolom itself was started.
AI coding assistantsApp security basicsNo-code and low-code
Building with an AI agent, one step at a time
An AI coding agent can build a working site from your words, but only if you steer it in small, checkable steps. This lesson starts a real Next.js project with the agent, gives it a briefing file, and shows the loop you will repeat for every feature.
AI coding assistantsPrompting for codeNo-code and low-code
Save points with Git and GitHub
Every working step deserves a save point you can return to, and a copy that lives somewhere safer than your laptop. This lesson connects your project to GitHub, shows you how to commit and push without typing Git commands, and how to undo a bad change in a minute.
Git and GitHubAI coding assistantsApp security basics
A database and sign-in with Supabase
Mangrove Kitchen shows recipes but can't remember anyone yet. Here you add a Supabase database and e-mail sign-in, so members can save favourite recipes, with every table locked by row-level security.
SupabaseApp security basicsAI coding assistantsPrompting for code
Going live on Vercel
Today Mangrove Kitchen leaves your laptop. You connect the GitHub repository to Vercel, give it the environment variables, point Supabase sign-in at the live address, and practise the two moves every owner needs: redeploy and roll back.
DeploymentApp security basicsSupabaseGit and GitHub
Capstone: Launch Mangrove Kitchen (or your own site)
A realistic project with a rubric and a model answer.
Final exam and certificate
10 questions drawn at random, 70% to pass.
Capstone project
Launch Mangrove Kitchen (or your own site)
Imran at Mangrove Market wants a small recipes-and-offers site, Mangrove Kitchen, to test whether recipes built around the weekly offers bring people back. Visitors read recipes; members sign in with e-mail, save favourite recipes and tick a box to get the weekly offers e-mail later. Build it, or your own idea of the same size, with an AI coding agent: a Next.js site in a GitHub repository, a Supabase database with sign-in and row-level security on every table, and a live deployment on Vercel. You write no code by hand, but you check every step: commit after each working step, test sign-in and access rules as two different users on the live site, and keep every secret key out of the code and the browser.
You will hand in
- A live site on Vercel (a vercel.app address is fine) where a visitor can sign up, sign in and save a favourite
- A public GitHub repository with a README, the SQL migrations the agent wrote, and at least eight small commits
- A build log: the instructions you gave the agent for three steps, what it changed and one thing you had to correct
- A safety check: which keys exist, where each one lives, how row-level security was tested, and that two-factor sign-in is on for every account
Check what you know
Already know some of this? Answer 5 quick questions, one from each lesson, and we will mark the lessons you can skim. It takes about 3 minutes and nothing is graded.
Questions
Do I really not need to code?
You don't write code by hand. You need to read a little: what files the agent changed, what a SQL statement creates, what an error says. Each lesson shows exactly what to look for. If you want more practice reading AI-written code, take Vibe Coding first.
Was Kolom really built this way?
Yes. Kolom was built by its owner working with an AI coding agent, on the same services this course uses: GitHub, Vercel, Supabase, Cloudflare, Resend and Google. The course teaches that process in small steps with a smaller site.
What will it cost?
You can finish this course on free tiers: GitHub, Vercel's Hobby plan and Supabase's free plan. The AI coding agent usually needs a paid plan or pay-as-you-go credits, and limits and prices change often, so check each provider's pricing page before you start. Most of these services bill international cards; if you live in Bangladesh, ask your bank about a dual-currency card enabled for international online payments.
Which AI agent should I use?
Any agent that can read and change files in a project folder and run commands works, such as Claude Code, Codex or the agent modes in AI code editors. The lessons use Claude Code as the example and point out what is the same in the others.
Is there an exam and a certificate?
Yes, and both are free. The exam mixes scenario questions with practical tasks such as putting launch steps in order or matching a problem to the right setting. You pass with 70% on the questions and an average of 70% on the tasks.
Is it really free?
Yes. Every lesson, podcast episode, exercise, quiz and final exam is free, so you can finish the course without paying. The labs and their templates for real work are free too. Plus and Premium add the verifiable certificate, every lab challenge, weekly extras and a mentor on top, and they never lock the tools or the courses.
How long does it take?
Each lesson shows an honest time for reading and doing. Most people finish a course in two to four weeks at three or four hours a week.