Everything we've covered in this bootcamp has pointed toward one thing: actually building something and publishing it.
That's what it's about. Not understanding every concept in theory, but going from "I have an idea" to a URL that actually works. It's possible to do in a single session. This episode gives you the map.
Step 0: Know what you're building
Before you write a line of code, write a sentence.
Not "a cool app." Not "basically like Airbnb but for X." A sentence that describes what a specific person should be able to do. For example: "A web app where users can leave feedback on products, see others' feedback, and log in with their email."
That sounds trivial. It isn't. Vague briefs give vague results — whether you're asking Claude for help or writing the code yourself. The precision of your description determines the quality of what you build.
Step 1: Set up the foundation
You need four things in place: a Next.js project, a Supabase project, a GitHub repo, and a connection to Vercel.
Claude Code can guide you through this if you describe the stack. Alternatively, do it manually in about ten minutes: `npx create-next-app@latest` creates the project, `git init` and a GitHub repo sets up version control, the Supabase dashboard gives you your keys, and `vercel link` connects the project to your Vercel account.
Put the Supabase keys in `.env.local`. Don't tell Git that file exists.
Step 2: Build with Claude Code
Open the terminal. Start Claude Code. Describe your app piece by piece, not as one enormous brief.
Start with the static. A homepage that shows products from Supabase. Then a product detail page. Then a form. Then login. Then styling. Then edge cases: what happens if there are no products? If a product doesn't exist (404)? If the network is down?
Each step is a separate conversation. Each conversation is verifiable. You see whether it works before moving on.
Step 3: Handle what doesn't work
Errors aren't a sign that the process is going wrong. They're part of the process.
When something doesn't work, describe exactly what you see: the error, what you tried, what you expected. Claude Code debugs, it doesn't just suggest. It reads the error message, examines the code, and iterates. It's faster and more systematic than searching Stack Overflow.
Step 4: Publish
Add your environment variables to Vercel (Supabase URL and anon key), run `vercel --prod`, and your app is live at a real URL.
Share it. Someone else being able to click on it changes how you think about the project.
The mindset that changes everything
The old way of building an app: learn the technology, build offline for months, publish when it's finished. The problem is that "finished" never arrived.
The new way: publish the simplest possible version of the idea as quickly as possible. Get feedback. Iterate. What works, you keep. What doesn't, you rewrite.
That's not a compromise on quality. It's the realisation that you don't know what quality means until you've seen how it's actually used.
The best time to build your first app was yesterday. The second best time is now.