Back when Django came out, it changed the game for backend development, admin panels, auth, and ORM were just… there. Then, Create React App made frontend development as easy as npx.
Now, with LLMs becoming more agentic, we’re exploring:
What if AI could build entire SaaS apps — from scratch — autonomously?
As a side project, we’ve been building SaaS-Builder, and we just opened-sourced it.
It’s an AI-native CLI framework that scaffolds full-stack apps using LLMs like GPT-4.1, Claude, or Gemini.
What it does:
You run:
saas-builder init
And behind the scenes, multiple agents kick in:
🖼️ UI Agent → generates a fully responsive Next.js frontend
🔐 Auth Agent → wires up Supabase Auth
🗃️ DB Agent → creates schema, migrations, types
🧪 Terminal Agent → installs packages, sets up folders
🌐 Browser Agent → launches the app locally
You describe your app → You get a clean, typed, ready-to-deploy SaaS scaffold with Supabase + Next.js.
It’s open-source (MIT), installable via pip, and still in early stages, but already usable.
Why we built this:
We wanted to stop repeating boilerplate every time we start a new project.
But more importantly, we wanted to see what agentic software development actually looks like.
If this idea resonates, try it, break it, suggest features, or just tell us if this direction makes sense.