๐ Inspiration
The idea for MakeAPP AI came from the common struggle developers, especially beginners and freelancers, face when building full web apps quickly. Creating a responsive, production-ready codebase from scratch takes time and expertise. With the rise of AI coding assistants, we saw the opportunity to build a tool that transforms a simple natural language prompt into a complete, styled web application โ instantly.
๐ What it does
MakeAPP AI generates a fully functional Next.js 14 App Router + Tailwind CSS project based on a single user prompt. It automatically creates:
A responsive, custom UI
MainComponent.tsx with layout and design
All essential files: tailwind.config.js, next.config.js, package.json, etc.
Clean code with no external class merging libraries (clsx, tailwind-merge)
Production-ready folder structure and live preview
Users can describe anything from a "portfolio website for a doctor" to a "SaaS landing page," and MakeAPP AI delivers the full code instantly.
๐ ๏ธ How we built it
We used:
Next.js 14 (App Router)
TypeScript
Tailwind CSS
Shadcn/UI (optional for future UI components)
OpenAI GPT-4 API for prompt-based code generation
Vercel for deployment
Sanity.io for structured content (coming soon)
Monaco or Sandpack for live preview
Edge Functions for handling streamed AI responses
The entire app runs client-side, but uses backend API routes to interact with OpenAI securely. User prompts are parsed, enhanced, and passed to a custom GPT-4 system prompt that ensures a consistent and reliable code structure.
๐งฑ Challenges we ran into
Dynamic iframe sizing: The Sandpack preview would not fully occupy the intended height, leading to issues with layout consistency.
Tailwind and dark mode clashes: Ensuring font and background colors contrasted properly across light/dark themes was tricky.
Prompt engineering: Getting the AI to generate complete and valid Next.js projects without clsx/tailwind-merge needed a lot of prompt refinement.
Live preview bugs: Some previews wouldnโt render correctly due to iframe or hydration issues.
๐ Accomplishments that we're proud of
Building a zero-to-app platform powered only by AI and a single user prompt
Creating consistent, production-grade code with minimal hallucinations
Making frontend web development accessible for non-devs and junior devs
Supporting multiple use cases: portfolios, doctor websites, business sites, landing pages
๐ What we learned
Prompt design is as important as UI/UX when building AI tools.
Real-time code preview tools require deep debugging and layout testing.
Small styling bugs can break UX and hurt trust in the product.
AI can be used not just to generate code, but to empower creativity.
๐ฎ What's next for MakeAPP AI
Team workspace support (collab on generated code)
Sanity CMS integration for real content management
Custom image generation via DALLยทE or Leonardo AI
Dark/light toggle UI
Backend + DB generation (Mongo/Postgres + Prisma)
Authentication support with Clerk or Auth.js
Export to GitHub, CodeSandbox, or Vercel in one click
Chrome extension for instant code snippets
Built With
- next.js
- openai
- sanity
- tailwind
- typescript
Log in or sign up for Devpost to join the conversation.