Day 3 - Build Your First App

Stop Planning. Start Building.

You've done the research. You've talked to users. You've written your PRD and killed your feature list down to one thing.

Now you build.

Here's the truth most people won't tell you: you don't need to know how to code. Not anymore. The tools available today let you describe what you want in plain English and get a working app back. It's not perfect. It's not production-grade on the first try. But it's real, and it's fast, and it's enough to test your idea with actual humans.

The goal today is simple: get something running in under 60 minutes.

Not polished. Not perfect. Running.

The Toolkit

There are three categories of tools. Pick the one that matches your skill level.

Build Full Apps From English

These tools take a description and generate an entire working app - frontend, backend, database, deployment.

• Lovable - Best for beginners. Describe your app, get a working prototype with a shareable URL. Great UI out of the box.
• Bolt.new - Fast prototyping. Good for quick experiments. Runs in the browser.
• Replit Agent - Best when you need backend logic, databases, or APIs. It handles the full stack.

Code With an AI Copilot

These tools help you write code yourself, but AI does the heavy lifting. You need some comfort with code, but not much.

• Cursor - VS Code with AI superpowers. You describe what you want, it writes the code. You review and edit. Best balance of control and speed.
• Claude Code - Terminal-based. For power users who live in the command line. Excellent at complex, multi-file changes.
• GitHub Copilot - Autocomplete on steroids. Good if you already use VS Code.

Generate UI From Descriptions

• v0 by Vercel - Describe a UI in words (or paste a screenshot), get React components back. Great for building individual screens fast.

The Decision Tree

Not sure which to pick? Here's the shortcut:

• Never written code before? Start with Lovable.
• Want more control over the code? Use Cursor.
• Just need a beautiful UI component? Use v0.
• Need a database and API? Use Replit Agent.

Pick one. Don't research for 3 hours. Just pick one and start.

The 60-Minute Sprint

Set a timer. Seriously. Put 60 minutes on your phone. When it goes off, you stop building and start showing.

Step 1 - Describe Your App in 3 Sentences

Before you touch any tool, write this down:

1. Who it's for. "Small landlords with 5-20 rental units."
2. What it does. "Automatically triages tenant maintenance requests and contacts the right vendor."
3. What makes it different. "The landlord doesn't have to do anything - no app to check, no messages to forward."

This is what you'll paste into your builder tool.

Step 2 - Use the GCES Framework

When prompting your AI builder, structure your request with GCES:

• Goal - What you want built. "A web app that lets tenants submit maintenance requests via text message."
• Context - Who uses it and why. "Small landlords who manage properties themselves and lose time playing phone tag with tenants and contractors."
• Examples - What it should look and feel like. "Similar to a simplified Zendesk ticket system, but for property maintenance."
• Steps - Break it into phases. "Phase 1: Landing page. Phase 2: Request submission form. Phase 3: AI categorization and vendor matching."

Step 3 - Build in This Order

Don't build everything at once. Follow this sequence:

1. Landing page. What does a visitor see? What's the headline? What's the call to action?
2. Core feature screen. The ONE thing your app does. Build this screen and make it work.
3. The happy path. Connect landing page to core feature. Make sure a user can go from "I just arrived" to "I got value" without hitting a dead end.

Step 4 - Test, Break, Fix

Open your app. Try to use it as a real user would. Click every button. Enter weird data. Try to break it.

When it breaks (it will), copy the error message and paste it back to your AI tool. Say: "I got this error when I clicked the submit button. Fix it."

That's the loop. Build. Test. Break. Fix. Repeat.

Don't Just Say "Build Me an App"

The biggest mistake beginners make is pasting a vague description and hoping for the best. You'll get generic slop.

Instead, use the architect pattern. Start with structure, then build piece by piece.

Bad prompt:

Build me an app for landlords to manage maintenance requests.

Good prompt:

You are a senior software architect. I want to build a web app for small
landlords to manage tenant maintenance requests.

First, outline the file structure and key components. Include:
- A landing page with value proposition
- A tenant request form (text + photo upload)
- An AI categorization engine that sorts requests by urgency and trade type
- A dashboard for the landlord to see all requests

Don't write any code yet. Just give me the architecture overview.

Once you have the architecture, build one component at a time:

Now implement the tenant request form. Use a clean, simple design.
Fields: description (text), photo upload, unit number, urgency (dropdown).
Make it mobile-friendly.

This approach gives you dramatically better results because the AI thinks about the whole system before writing code.

Things That Will Waste Your Time

Building 10 features instead of 1. You crossed out 4 features yesterday. Don't sneak them back in. Build the one thing.

Not testing with a real user in the first 24 hours. Your app doesn't need to be pretty. It needs to be in front of a real person who has the problem you're solving. Send them the link tonight.

Spending 3 days on the logo. Your logo doesn't matter. Your color scheme doesn't matter (yet). Whether your app solves a real problem matters. Everything else is procrastination disguised as productivity.

Rewriting from scratch instead of iterating. When something doesn't work, fix it. Don't start over. The code doesn't need to be perfect. It needs to work.

Not reading the error messages. When your build fails, the error message usually tells you exactly what's wrong. Read it. Paste it into your AI tool. Let it fix the problem.

Your Day 3 Assignment

The Build Sprint

1. Open your tool (Lovable, Cursor, Bolt, whatever you picked)
2. Set a 60-minute timer
3. Build: landing page + core feature + happy path
4. When the timer goes off, stop
5. Send the link to one person who has the problem your app solves
6. Write down their reaction in 2 sentences

The goal isn't a finished product. The goal is a working prototype that a real person can try. Ship ugly. Ship broken. Just ship.