From Concept to MVP in 2 Days: My Vibe Coding Experience in Building a Yelp-like app

Two days.

That’s how long it took to go from “I have an idea” to a deployed MVP that real people could actually use.

That still feels wild to say out loud—especially when I compare it to the six months it took me to build my last web app.

This post is a reflection on how modern tooling + AI + the right mindset completely changed the way I build software, using my latest project, Little Weavers, as the case study.

Step 1: Asking My Wife for an Idea (The Best Product Move I Made)

I didn’t start with market research or trend analysis.

I started by asking my wife a simple question:

“If I were to build an app, what would actually be useful?”

She immediately talked about something close to home: how isolating motherhood can be, especially for new parents. Finding mom groups, playdates, events, or even online communities often means digging through Facebook groups, outdated websites, or word-of-mouth.

That conversation became the seed for Little Weavers.

Step 2: Defining the Problem — Combating Isolation for New Parents

The core idea crystallized quickly:

Parenting (especially early on) can feel lonely
Moms want connection—for themselves and their kids
The information already exists, but it’s scattered

Little Weavers’ mission became simple:

Create a single place where mothers can discover groups, meetups, and events—both online and offline—tailored to their location and interests.

Once the “why” was clear, everything else flowed faster.

Step 3: Letting ChatGPT Help Choose the Tech Stack

Instead of overthinking architecture for days, I leaned into something new: asking ChatGPT directly.

I explained:

My experience level
The kind of app I wanted (location-based, auth, database-driven)
My desire to move fast, not build a perfect system

From there, ChatGPT recommended:

React + Next.js for the frontend
Supabase for database, auth, and backend speed
A lightweight MVP-first approach instead of premature scaling

This alone probably saved me a week of decision fatigue.

Step 4: MVP Feature Definition (In Minutes, Not Meetings)

Next, I used ChatGPT to define the MVP feature set.

Instead of bloated roadmaps, we focused on:

Browsing groups and events
Location-based filtering
Categories (online, offline, interests)
User accounts
Saving/favoriting items

Clear, achievable, and shippable.

No fluff.

Step 5: Database Schema + Supabase = Turbo Mode

I then asked ChatGPT to help design the database schema:

Users
Events
Groups
Locations
Saved items

From there, Supabase completely changed the game.

In a fraction of the time it used to take me:

Auth was set up
Tables were created
Relationships were defined
Row-level security was handled

What used to be weeks of backend wrangling became an afternoon.

Step 6: Designing a Working Prototype with Figma Make

This was one of the biggest “wow” moments.

Using Figma Make, I created a working prototype, not just static screens. I could click through flows, test layouts, and validate UX decisions before writing serious code.

Even better:
I could download the generated code and use it as a real starting point.

Design and engineering finally felt like one continuous process.

Step 7: React + Next.js + Figma Code Integration

I pulled the Figma Make code into a React / Next.js project and started wiring things up.

Because the UI already existed:

I wasn’t staring at a blank screen
Components had structure
Styling decisions were mostly done

Instead of fighting CSS, I focused on making things work.

Step 8: GitHub Copilot as My Pair Programmer

This is where things felt almost unfair—in a good way.

With GitHub Copilot, I:

Replaced mock data with real Supabase queries
Solved bugs by describing intent instead of Googling
Generated SQL when I needed schema changes
Iterated faster without breaking flow

It felt less like coding alone and more like collaborating with an extremely fast junior dev who never gets tired.

Step 9: Deployment in Minutes with Railway

Once the app was usable, deployment was almost anticlimactic.

I connected Railway to my GitHub repo, clicked deploy, and… that was it.

No multi-day DevOps spiral.
No cryptic config hell.

Just:

Repo → Build → Live URL

Seeing a real app online two days after the initial idea felt surreal.

Looking Back: From 6 Months to 2 Days

The last time I built a full web app, it took over six months.

That project—Portland Patent—allowed users to submit provisional patents online, sign documents virtually, and skip endless email back-and-forth with attorneys. It was powerful, but building it meant:

Learning Firebase from scratch
Manually wiring auth, storage, rules, and flows
Constant context switching
Slow iteration

Little Weavers, by contrast, came together in 48 hours.

Not because it’s simpler—but because the ecosystem has changed.

Final Thoughts: This Is a New Era of Building

This experience fundamentally shifted how I think about software development.

Today:

AI helps you think, plan, and execute
Backend platforms remove unnecessary friction
Design tools generate usable code
Deployment is no longer a blocker

What matters most now isn’t knowing everything—
it’s knowing how to move, how to ask the right questions, and how to stay in flow.

If this is what “vibe coding” looks like, I’m all in.

And Little Weavers is just getting started. 💛

From Concept to MVP in 2 days: My Vibe Coding Experience in Building a Yelp-like app