Ship

Get it live

Fix Things and Keep Going

Things will break. Here is the minimum you need to keep moving.

You have real users now. Things are going to break -- that's normal. This guide gives you the minimum knowledge to fix problems, keep your project organized, and keep improving.

A note on where this comes from. The Army developed the After Action Review at the Combat Training Centers -- where I spent a year as an Observer/Controller/Trainer, training units in running AARs. The AAR is three questions: what was supposed to happen, what actually happened, what do we do differently next time. No rank in the room. I watched battalion commanders with 20 years of experience realize in the AAR that their plan had fallen apart and they hadn't seen it. The realistic training showed them the cost without real casualties. The weekly retro at the end of this guide is the same mechanism, applied to building. It forces you to say "I spent 8 hours on something that didn't matter" before you waste another 8.

Three Things That Make Up Every Website

Every website or app -- no matter how complex -- is built from three things. AI writes all three for you. You just need to know which one is causing a problem.

What It's Called	What It Does	When Something's Wrong, You'll See
HTML	The content and structure. Text, images, buttons, forms -- what's on the page.	Something is missing, not showing up, or in the wrong place
CSS	How it looks. Colors, spacing, fonts, layout -- the visual design.	Things overlapping, wrong colors, looks broken on your phone
JavaScript	What it does. Clicking buttons, loading data, saving information.	Nothing happens when you click, data doesn't load, error messages

You don't need to understand the code. You need to understand which category the problem is in, then tell your AI: "The button exists but nothing happens when I click it" (that's a JavaScript problem) or "The text is showing up but it's overlapping the image on mobile" (that's a CSS problem). The more specific you are, the better the AI's fix will be.

How to Fix Things When They Break

This is the single most important skill. Not writing code -- copying error messages and giving them to your AI with context.

Step 1: Open your browser's developer tools

Right-click anywhere on your page and choose "Inspect" (Chrome/Edge) or "Inspect Element" (Firefox/Safari). This opens a panel that shows you what's happening behind the scenes.

Step 2: Check the Console tab

Click the "Console" tab. Red text = errors. This is where you'll find out what went wrong. Copy the red text exactly -- don't try to interpret it yourself.

Step 3: Give it to your AI with context

Paste the error message and explain what you were trying to do:

"I clicked the Submit button and got this error: [paste the red text]"
"The page loads but the list of items is empty. Here's the console error: [paste]"
"It works fine on my computer but when I open it on my phone, the layout is broken"

That's 90% of debugging. Read the error, copy it, tell the AI what you were doing when it happened. The AI will almost always know how to fix it. You don't need to understand the fix -- you need to know how to find the error.

Common Problems and Where to Look

What's Happening	Where to Look	What to Tell Your AI
Page is blank or broken	Console tab -- look for red errors	Copy the exact error message and paste it
Layout looks wrong	Try resizing your browser window	"The [thing] is overlapping [other thing] on mobile"
Button does nothing when clicked	Console tab -- look for errors after clicking	"I click [button] and nothing happens. Console shows: [paste]"
Data isn't showing up	Console + Network tab	"The page loads but the [data] is missing. Error: [paste]"
Works on your computer, breaks online	Hosting dashboard + deploy logs	"It works locally but not when I deploy. Here's the error: [paste]"

Warning Signs in AI-Generated Code

You can't read every line the AI writes. But you can spot these red flags:

Passwords or secret keys visible in the code -- these should never be in your code files. If you see something that looks like a long random string, ask your AI: "Is this a secret? Should it be in an environment variable instead?"
Fake data that looks real -- AI loves to generate placeholder names, emails, and products. Make sure your app is actually connected to real data, not just showing demo content.
No "loading" or "error" states -- if your app loads data from the internet, something should show while it's loading. A blank screen makes people leave. Tell your AI: "Add a loading message while the data loads, and an error message if it fails."

Keeping Your Project Organized

Builder's checkIf you're about to set up a project management tool, stop. I watched myself do this -- spent an afternoon configuring a board with columns, writing tickets with acceptance criteria, tagging things by priority. For a team of one. Nobody was ever going to read any of it. PM tooling solves coordination problems. You don't have a coordination problem. You have a building problem. A text file with three bullets about what you're doing next and a flat list of tasks is the entire system. If your "project management" takes more than ten minutes a week to maintain, you're not being disciplined. You're avoiding the work. And when you sit down to build, ask yourself: is this urgent or is it important? Fixing the bug a user reported is urgent. Talking to three more potential customers is important. Most solo builders spend all their time on urgent and never touch important. The urgent will always be there. The important is what decides whether you're still here in six months. More on this.

After a few sessions with AI, your project can get messy. Files everywhere, duplicated code, things that don't work anymore. These habits prevent that:

1. One file should do one thing

If you're not sure what a file does from its name, that's a problem. Tell your AI: "This file is getting big. Split it into smaller pieces and name them clearly."

2. Tell the AI about your existing project

Before each session, give the AI context: "Here's my project. I have files for [X, Y, Z]. Follow the patterns that already exist. Don't create new folders or reorganize things." This prevents the AI from reinventing your project structure every time.

3. Delete what you're not using

Old code that's not being used confuses both you and the AI. If you stopped using something, delete it. If you're worried about losing it, that's what version control is for (ask your AI: "Help me set up Git so I can undo changes if needed").

4. When in doubt, ask the AI to explain

You can always say: "Explain what this file does in simple terms" or "What would happen if I deleted this?" The AI is a patient teacher -- use it.

As Your Project Grows

These matter later -- not now. Come back to this section when you have real users.

When	What to Do
0-10 users	Don't worry about code quality. Get it working and get it in front of people.
10-50 users	Keep files small and named clearly. Ask your AI to clean up anything confusing.
50-100 users	Ask your AI about TypeScript (catches mistakes automatically) and testing (makes sure payments work).
100+ users	Invest in proper structure. At this point, consider hiring a developer for a few hours to review your code.

The test: Can you make a change to your app in 20 minutes? If yes, your code is fine. If every change takes hours and breaks other things, ask your AI to help you reorganize.

Go deeper: The State of the Solo Builder: 2026 — data on burnout rates, sustainability, and the human cost of building alone.

Grow

Find Your First Users