It kept asking for more (sign-up, notifications, permissions)0
I just forgot it existed0
Honestly? No idea. It just didn't stick0
Before you touch Claude Code
Every app on your phone started exactly like yours will
Not with code. With a decision about what to leave out.
You first
How many features did Weixin (WeChat) launch with, in 2011?
▸ click to reveal after they answer
One. Text messaging. No Moments, no Wallet, no Mini Programs, no mini-games. Those all came years later.
The loop
Every product you use was built like this
Spec
The smallest thing worth making
Build + ship
Hand it to strangers
Iterate
Watch what people actually do
It never stops. You are about to run one full lap, today.
One company actually lived this
2010, San Francisco.
Real example
Let's walk one real app through the whole loop: Instagram
2010, San Francisco
The app before the app
Burbn, the spec
Check in to places. Make plans with friends. Earn points for hanging out. Post photos. Message people. Basically everything.
Instagram, what shipped
Take a photo. Add a filter. Share it. That's it.
Site walkthrough · giants are born ugly
Two more that started out looking like nothing.
versionmuseum.com/history-of/amazon-website
amazon.com
Earth's biggest bookstore
Books | Search | Browse Subjects | Bestsellers gray background, plain blue links, zero images above the fold
Amazon, 1995
versionmuseum.com/history-of/google-search
BackRub
a Stanford class project, one text box, no logo yet
Google, 1996 (before it was even called Google)
Real screenshots, verified via versionmuseum.com. Ugly first, huge later. Every time.
Step 3, years later
Stories, Reels, ads, DMs: all came after, not before
None of that existed on day one. It got added because millions of people were already using the ONE thing, and Instagram watched what they wanted next.
The lesson
The loop never stops. Not even for Instagram.
Your job today is one full lap: spec, then ship. That starts with getting scope right.
So, what's an MVP?
MVP: the smallest version that proves your idea works
Minimum Viable Product. Not minimum EFFORT, minimum SCOPE. It still has to actually work, just at the smallest size that teaches you something real.
Pick one to build this week
The everything-app vs. the one-thing-app
The everything-app
StudyBuddy Ultra: flashcards + forums + calendar + AI tutor + group chat + leaderboard + dark mode + a store where you buy hats for your avatar
The one-thing-app
Paste your class notes. Get 5 quiz questions back. That's the whole app.
You first
What was Instagram called before it was Instagram?
▸ click to reveal after they answer
Burbn. The everything-app from three slides ago. It almost died bloated before it got cut down to one thing.
Two words you will use all week
Spec
A one-page plan of what you are building. Who it is for, the ONE thing it does, what done means.
Deploy
Putting your product on the internet so anyone with the link can use it. Deployed = real.
The enemy is scope
Cut until it hurts. Then stop.
Scope kills more student startups than competition ever will. Great Day-5 products do ONE thing a stranger can use, without help.
Exercise 1
Now you do it: the knife
Exercise 1 · why this nowOFFLINE · screens closed
Paper first. No laptops.
You cannot cut what you have not written down. List everything your product could do BEFORE you touch a keyboard to remove anything.
Destination: THE WALL, printed after class
Exercise 1 · do it nowOFFLINE · screens closed
The knife
6:00 on the clock
1List everything your product COULD do. Everything.
2Cross out lines until removing one more kills the point.
3What survives = your ONE feature. Write it as: 'A stranger can [action] at our URL without our help.'
6:00
Then: share with the class
Exercise 1 · debriefOFFLINE · screens closed
What survives a good cut
Before the knife
"IB study app": flashcards + forums + calendar + AI tutor + past-paper library + study groups
After the knife
A stranger can paste their class notes and get 5 quiz questions back.
You first
How do you know your ONE feature is actually done?
▸ click to reveal after they answer
Not when you think it's done. When someone who has never met you can use it alone.
The done-definition
“A stranger can [action] at our URL. Without our help.”
If you have to stand next to them and explain it, it is not done. It is a demo.
Two specs
Which one actually tells you when to stop building?
The vague spec
An app that helps students study better using AI
The stranger-can spec
A stranger can paste 10 vocab words and get a 5-question quiz back, at our URL, with zero instructions from us.
Assemble it first
Now write the actual one-page spec.
Three parts: your thesis at the top, your ONE feature from the knife, and what done means ("a stranger can [action] at our URL without our help"). Ask Claude to draft it from your knife survivors; you edit. Destination: TEAM FOLDER.
Copy this prompt · run it now
Spec attack
Here is my product idea and my one-page spec draft. Attack it: what is too big for a 4-day build? (Notice: context, instruction, format. Day 1 anatomy, bigger stakes.) What would you cut? What breaks first?
Time to build, for real
The spec becomes a page.
Mission 0 FOLLOW ALONG · laptops open
Builder, get the team in. Before anything else.
3:00
1Builder opens claude.ai/code (or the desktop app) on the team laptop.
2Log in with the TEAM account card on your table, not a personal one.
3Every teammate confirms they can see the screen. Hands up when your whole team is in.
3:00
Every team in. Then we watch it work.
First contact
Meet Claude Code
What an agent is
It doesn't just describe the code. It writes it, runs it, ships it.
You have used chatbots that talk about what code could do. Claude Code opens your files, writes real code, runs it, and puts it on the internet. This is the tool your whole team uses for the rest of the week.
Site walkthrough · people just like you, already shipped
A habit tracker. A trip planner. A link shortener. All real, all AI-built.
lovable.dev/templates
Continuum
A calm habit tracker with streaks
Meeting Booker
Personal scheduling, made simple
URL Shortener
Shorten and track any link
Real, free site: lovable.dev/templates. One person, one AI, one weekend. Your team ships in 30 minutes.
Quick vocabulary
A terminal is just texting your computer without the pretty buttons.
Programmers love it because it is fast. You will see me use it for ten seconds to look cool. You will use Claude Code as a normal app, in a window, like everything else. Nobody needs the scary black screen this week.
First-contact demo · live, no rehearsal
Our spec, live, on the internet before the break.
LIVE DEMO · no rehearsal
"Build this one-page spec, deploy it, give me the live URL."
Real spec, real deploy, live. If it breaks, it breaks live. Nothing about this is magic.
Just watch
Watch the terminal, not your own laptop.
Notice it reads files before it writes any.
Notice it explains its plan before acting.
Watch for the moment a real URL appears.
Your turn is next: Exercise 2, laptops open then.
The save-point rule
Commit before you experiment.
Save a version first. Then try anything. You can always get back to where you were. Fear of breaking it disappears today.
Exercise 2
Your turn: ship Hello World
Exercise 2 · why this nowFOLLOW ALONG · laptops open
Right now, "putting something on the internet" feels hard. In 30 minutes it won't.
That belief change is the whole point of today. Everything you attempt for the rest of the week depends on it being gone.
Destination: CLASS CHANNEL, every URL posted
Copy this prompt · run it now
The exact instruction to give Claude Code
Build a one-page site for our team. It needs our team name, our one-sentence thesis, and all our names. Make it look clean, deploy it, and give me the live URL.
This is what you're about to ship
Your URL, in thirty minutes
teamname.vercel.app
Team Nightowl
A stranger can paste their class notes and get 5 quiz questions back.
Built by Aiden, Lucy, and Marco
Try it →
Exercise 2 · do it nowFOLLOW ALONG · laptops open
Ship Hello World
30:00 on the clock
IF YOU GET STUCK
Stuck 3+ minutes? Raise a hand, keep trying. Still nothing at 20:00? Switch to Lovable, same instruction, no shame. Live is live.
Ask yourself: what is the URL supposed to show?
That question gets you unstuck faster than the fix itself.
You do, step by step
1Open Claude Code with your team's project.
2Give it the instruction from the last slide. Deploy it.
3When the URL loads on YOUR phone, commit before you experiment.
4Post the URL in the class channel.
30:00
URLs on the big screen
If you get stuck
Don't say "it's broken." Say this instead.
Gets you nowhere
it's not working. fix it??
Gets you unstuck
The page is blank when I open the URL. It should show our team name and thesis instead. Check what's wrong and fix it.
Checkpoint · 16:15
Your names. On the internet. Day one.
Tomorrow: real customers.
“Make something people want.” Today you proved it.
Go deeper · on your own time
Four places to keep exploring this
Lovable templates
A gallery of real, simple apps other people built with AI, browse for ideas.
lovable.dev/templates
Amazon & Google's ugly first versions
Real, restored screenshots of the giants at your exact starting point.
versionmuseum.com
"How we Claude Code"
For the ambitious: watch Anthropic's own engineers use the tool day to day.
youtube.com/@claude
Claude Code docs
The official overview page. Day 3 goes deeper, this is a preview.
code.claude.com/docs
Next: 16:15
Standup: 60 seconds per team, thesis + URL on the big screen
Then tomorrow, you leave the building: real interviews with real strangers, starting with your own family tonight.