Skip to main content
Week 01 • Track 03 • The Workflow

The Translation
Layer.

You don't need to speak Python. You need to speak Logic.
The "Studio Flow" is the art of translating a messy human idea into a rigid document the machine can respect.

ABSTRACT IDEA
SPEC FILE
WORKING APP

Bridging the Gap

Ambiguity is the Enemy.

When you say "Make it pop," the AI hallucinates. When you say "Increase border-radius to 12px," the AI executes.

The "Studio Flow" requires you to stop being an Idea Guy and start being an Architect. You must create an intermediary file—the Spec—before a single line of code is written.

"A login page"
Auth Form (Email/Pass) + Supabase
"Like Instagram"
Feed Component + Infinite Scroll
"Store data"
PostgreSQL Table: 'users'

The Blueprint Process

🧠

1. The Brain Dump

Get the chaos out. Write everything you want the app to do in plain English. No technical terms yet.

👤

2. The User Story

Formalize it. "As a [User], I want to [Action], so that [Benefit]." This defines the scope.

💾

3. The Data Model

The most critical step. Define your Objects (Nouns) and their properties before designing UI.

.md

The Artifact: SPEC.md

This is the file you feed to Cursor/Replit.

# PROJECT: Habit Tracker MVP

# 1. CORE FLOW
- User lands on homepage (public).
- User signs up via Google Auth (Supabase).
- User sees a dashboard of "Habits" (checkboxes).
- User clicks checkbox → database updates.

# 2. DATA STRUCTURE
Table: users
  - id (uuid, primary key)
  - email (string)
  - avatar_url (string)

Table: habits
  - id (uuid)
  - user_id (ref → users.id)
  - title (string)
  - is_completed_today (boolean)

# 3. UI/UX
- Font: Inter
- Theme: Light mode only
- Components: Shadcn 'Card' for habits

💡 Tip: The more specific your Spec, the less the AI will hallucinate.

Spec Checklist

Before feeding the AI, verify:

  • Flow is linear (no branches)
  • Data types are explicit
  • Max 4 screens defined
  • Tech stack named
  • UI components specified
  • Auth method chosen
  • No "make it pop"
  • No "like Instagram"

The Spec Components

A complete spec has five essential sections. Missing one creates gaps the AI will fill with hallucinations.

📋

Project Overview

WHAT IS THIS?

One paragraph. What does it do? Who is it for? What problem does it solve?

// Example

"A habit tracker for busy professionals

who want to build consistent routines

without complex gamification."

🔄

User Flow

STEP BY STEP

The journey. Landing → Sign Up → Dashboard → Action → Result. Keep it linear for MVP.

// Example

1. User lands on homepage

2. User signs up (Google Auth)

3. User sees habit dashboard

4. User checks off habits

💾

Data Model

YOUR NOUNS

Tables and fields. Every Object (User, Post, Habit) needs explicit properties and types.

// Example

habits: id, user_id, title,

is_completed (boolean), created_at

🥞

Tech Stack

YOUR TOOLS

Be explicit. Framework, styling, database, auth, icons. No ambiguity.

// Example

Next.js 14, Tailwind, Supabase,

Shadcn/UI, Lucide Icons

Remember: The spec is a contract. If it's not in the spec, it doesn't exist. The AI will not read your mind.

Common Spec Mistakes

These mistakes create specs that confuse AI and produce broken apps.

Mistake #1
🌊

Scope Creep

"Also add notifications, social sharing, and a recommendation engine." MVP means ONE thing done well.

Too many features
Unclear priority
Bloated scope
Mistake #2
🫥

Vague Data

"Store user info." What info? Name? Email? Avatar? Address? The AI will guess wrong.

Missing field names
No data types
No relationships
The Fix
📐

Be Surgical

One core flow. Explicit fields. Named components. If you can't draw it, don't spec it.

3-4 screens max
Every field typed
Linear user flow

The Exercises

Practice the translation before touching code.

01

The Brain Dump

Get Ideas Out

  • Pick an app idea (Book Library)
  • Write everything in plain English
  • No technical terms allowed yet
  • List all features you want
Goal: Empty your head first
02

The Data Model

Define Your Nouns

  • What is a "Book"? List fields
  • title (string), author (string)
  • rating (1-5), finished (boolean)
  • What is a "User"? List fields
Goal: Know your data shapes
03

The Spec Review

Get AI Feedback

  • Write your SPEC.md file
  • Paste into ChatGPT/Claude
  • "Critique this for edge cases"
  • Refine based on feedback
Goal: Bulletproof your spec

📋 Quick Reference Cheatsheet

User Story

"As a [User], I want [Action]"

Data Field

field_name (type, constraints)

Flow Step

1. User does X → sees Y