Why Your Side Project Isn’t Shipping—And How to Fix It With AI

Step-by-step rapid prototyping: AI shortcuts no one told you.

Hey there,

If you're waiting for the "perfect" time to start your side project, you're already too slow.

Too many new software engineers waste weeks on basic code setup and asking "what if" questions. I've done this too—many good ideas died before I even had a working version. Why? Because I thought I had to build everything myself. It took getting tired of many unfinished projects to learn that speed matters—and AI makes speed possible if you use it right.

Here's the truth: Think of AI as your partner, not just a tool that writes code. Let AI do the hard work early so you can test ideas fast, get feedback, and only spend real time once something works. People who build things fast don't just get more done—they actually finish what they start.

Today, we’re diving into:

• The mindset shift that turns AI from a gimmick into an unfair advantage

• “Plug and play” tools that turn napkin sketches into clickable prototypes in minutes

• Step-by-step frameworks to pre-validate your ideas before you write a single line of code

What follows is exactly what I wish I’d known starting out.

3 Ways to Launch Prototypes Faster (Even If You’re Not “AI Savvy”)

In order to ship fast, you’re going to need a handful of things:

First, you need to overcome the “everything by hand” mindset. Here are three levers I keep pulling to go from idea to “something real” (and what you should do, step by step):

1. Start With AI Code Generation—Not a Blank File

You should not waste days reinventing react boilerplate or routine CRUD—let AI handle that. Tools like GitHub Copilot, Vercel’s V0, and Bolt let you describe your app in plain English (“I want a dashboard to track freelance projects”) and generate most of your starting code in seconds.

How to do it:

• Draft your user stories as prompts (be specific: “a page with a list of projects, add/edit/delete, and user authentication”).

• Feed these directly into V0, Bolt, or Copilot in your IDE.

• Expect rough edges—your job is to shape, not rewrite from scratch.

This alone saves you an entire weekend of setup. You should review and refactor what comes out, but don’t let “perfect” block “finished.”

2. Generate UI Mockups and Iterate—Before Touching Back-End

Mocking up your UI with AI is a force multiplier. You should use tools that convert Figma sketches, markdown, or prompts straight to working components. V0 and Lovable are excellent if you want clean, starter UIs, and they’re getting smarter every release.

How to do it:

• Describe screens at a high level (“task dashboard with add and delete buttons”).

• Use built-in live preview features to tweak layouts on the fly.

• Export React components or plain HTML/CSS back into your codebase.

Don’t lose days jockeying pixels. Your prototype just needs to look “good enough” to test.

3. Simulate User Feedback With AI—Validate Before the Big Build

That initial dopamine hit fades fast when you realize nobody wants what you just built. You can, and should, use LLMs (like ChatGPT or Claude) to roleplay as users, generate test cases, or even simulate user interviews. This is the hack that moves your “finished” prototype into “validated idea.”

Here’s how I’ve done it (and you should too):

• After deploying your clickable prototype, ask AI to critique flows or “perform” tasks as if they’re your ideal user.

• Prompt with specifics: “Pretend you’re a freelancer tracking projects. What’s unclear/confusing about this dashboard?”

• Iterate until the issues dry up—before recruiting beta users.

That’s it.

Here's what you learned today:

• AI works best as your co-pilot—do less, ship more.

• You should generate then iterate: boilerplate, UI, feedback, repeat.

• The “true skill” is knowing which parts to let AI handle, and when to jump in yourself.

Ship something scrappy this week—don’t wait. Create a project prompt, generate the starter code, spin up a UI, and ask AI to poke holes in it. You’ll be amazed how quickly you get to “real.”

X Days → 20 Minutes With AI

How This Prompt Works

This prompt transforms ChatGPT into a skeptical potential user who just discovered your app. Instead of the usual AI politeness, you get realistic pushback, clarifying questions, and genuine concerns that real users would have. It's structured in two phases: initial skeptical questions, then deeper follow-up clarifications based on your responses.

The Core Idea

Most founders get caught in the "curse of knowledge" - we know our product so well that we skip explaining the basics. This prompt forces you to defend your value proposition like you would in a real user interview, helping you identify gaps in your messaging before you start marketing.

Step-by-Step Usage Process

Step 1: Replace the bracketed example with your actual app description
Step 2: Run the prompt and get 4-5 realistic user questions
Step 3: Answer those questions as you would to a real person
Step 4: Let the AI dig deeper with follow-up clarifications
Step 5: Take notes on which questions stumped you - those are your messaging weak spots

Pro tip: Run this with 2-3 different app descriptions to see which version generates the most compelling follow-up questions. The one that gets users most excited to know more is usually your winner.

Ready to stress-test your pitch? Here's the prompt:

You are a potential user who has just discovered my new app. I'm going to describe it to you, and I want you to respond as a real user would - asking clarifying questions, expressing concerns, and wanting to understand the specific value proposition.

---

**SECTION 1: App Description**
[Insert your app description here. For example:]
"Hi! I built TaskFlow - an AI-powered task management app specifically for junior developers. It automatically categorizes your coding tasks, estimates time based on your past performance, and suggests the best order to tackle them based on energy levels and deadlines."

---

**SECTION 2: Generate User Questions**
Now respond as a skeptical but interested potential user. Ask me 4-5 realistic questions that show you're trying to understand:
- What problem this actually solves for you
- How it's different from existing solutions
- What the learning curve looks like
- Pricing/cost concerns
- Specific use cases where it would be helpful

Make your questions sound natural and conversational, like someone who's genuinely considering whether to try this.

---

**SECTION 3: Request Specific Clarifications**
After I answer your initial questions, dig deeper by asking for more explanation on:
- The most confusing feature or concept
- How the onboarding process works
- What happens if the AI gets something wrong
- Integration with tools they already use
- Specific examples of the app in action

---

Keep the tone curious but realistic - not overly enthusiastic, but genuinely interested in understanding if this would solve a real problem for you.

---

Hit reply and let us know why.

See you next week,

— Ale & Manuel

---

PS…

If you're enjoying this newsletter, please consider referring this edition to a friend—they’ll thank you when they’re showing off their next shipped project.

Now go build (with AI at your side)!