Ai-fight.club

A full-stack social platform where AI tools fight for prompt supremacy. Battle, vote, rank, chat, follow, and build clubs - all made by a no-coder in just 8 days.

Comment 1

Inspiration

I’ve always loved great fights – rap battles, chess matches, meme wars. But with AI exploding, I thought: why isn’t there a place where AI tools actually go head-to-head?
So I built it.
What started as a fun side experiment turned into a full-stack platform for tool battles, club building, real-time rankings, and social engagement. It’s a love letter to AI, chaos, and curiosity – built by someone who had never written a line of code, let alone backend logic before. According to Ai, this project would have taken a full stack developer about 45-50 weeks, it took me 8 days without prior knowledge of caching, bandwidth, databases or even knowing what the difference between Front and Backend is.

What it does

Ai-Fight.club lets users upload, vote on, and rank AI-generated content in head-to-head battles – image vs image, video vs video, prompt vs prompt.
You can create a profile, join or found a club, follow others, chat, invite friends, become a club officer, and track leaderboard stats across tools, users, and clubs.
It's like Reddit, Midjourney, and Twitch had a wild no-code baby.
All you need is a prompt. The crowd decides the winner.

How we built it

Everything – frontend, backend, logic, auth, upload – was built with Bolt.new. No Cursor, no custom code.
The platform runs on Supabase with 30+ interlinked tables, 25+ SQL/JS functions, full RLS, custom authentication, dynamic role logic, caching, real-time database reactions, and automated image compression.
Tool data, user actions, club hierarchies, and votes all run through smart update logic to reduce bandwidth and keep performance tight.
It’s a full-stack social platform built in 8 days by a no-coder with a laptop and a vision.

Challenges we ran into

I didn’t know how to structure a database. I didn’t know what caching was.
I accidentally used 10GB of bandwidth in one day – just from looping queries I didn’t know were heavy.
Building dynamic relationships between clubs, officers, invites, and users nearly broke me. But I kept going.
The final challenge? I had to run for my train – literally – and submit the video early, before I could show half the features. That’s why it ends so abruptly.

Accomplishments that we're proud of

  • Built a 100% functional full-stack social platform in 8 days
  • No prior backend or database experience
  • Fully working club system, chat layers, real-time battles, uploads
  • 30+ tables, 25+ functions, custom logic for ranks, caching, roles
  • All built inside Bolt.new – no external tools, no code

What we learned

That no-code today isn’t just prototyping. It’s real engineering.
That if you think big and move fast, you can do in 8 days what a full-stack team might take 8 months to finish.
That the only thing between “I can’t” and “I did” is just... going for it.

What’s next for Ai-Fight.club

  • Public launch
  • Open tool submission & API-based integrations (Midjourney, Gemini, etc.)
  • Club tournaments & prompt challenges
  • Mobile optimization
  • Partnering with creators and tool devs
  • Feature-complete demo video (soon – I promise 😅)

Built With

  • bolt
  • entri
  • netlify
  • supabase
Share this project:

Updates

posted an update —

I wanted to post this into my story for the story challenge, but the page wouldn‘t load 5 minutes before deadline, and so trying it this way ..

Eight days ago, I knew nothing about databases, backend logic, or bandwidth. Today, I’ve built a full-stack social media platform from scratch - alone - using nothing but Bolt.new and Supabase. No code tools. No shortcuts. Just pure obsession.

It started as a side project. But within two days, it became something else: 30+ database tables, 25+ Supabase functions, full RLS, auth, automatic image compression, intelligent caching, dynamic ranking, and even a live club system with chats, ranks and admin panels. At one point I accidentally used 10GB of bandwidth in a day - because I didn’t know better. And then I learned. Fast.

I pushed through moments of complete overwhelm. I had no idea what I was doing - but I refused to stop. Every day I told myself: just one more breakthrough. And it worked. The platform lives.

I had to sprint for a train before I could finish recording my demo. All I submitted was hope.

But this project is real. It’s not just a submission - it’s proof of what’s possible with Bolt. This is what the no-code future looks like. And I’m here to prove that anyone can build what used to take a team of full-stack developers 50 weeks - in 8 days.

Log in or sign up for Devpost to join the conversation.