AI-Powered Landing Page Personalizer

This project is a web application that personalizes landing pages based on the content of an ad.
The goal is to improve how well a landing page matches the message of an ad, which is important for user experience and conversions.

---

What the project does

The application takes:

• An ad (text, image, or URL)
• A landing page URL

It then:

1. Extracts key information from the ad (headline, tone, call-to-action, etc.)
2. Scrapes the landing page
3. Modifies parts of the page (like headline and buttons)
4. Generates a personalized version of the same page

The important part is that it keeps the original layout and only changes content where needed.

---

Tech Stack

• Backend: FastAPI (Python)
• Frontend: React (pre-built and served as static files)
• Web Scraping: Playwright
• HTML Parsing: BeautifulSoup
• AI (optional): Gemini API
• Fallback logic: Rule-based (used when AI is unavailable)
• Deployment: Render (Docker)

---

How it works

1. The user provides an ad and a website URL
2. The backend:
   • Analyzes the ad
   • Loads the full website using Playwright
   • Parses the HTML
3. The system updates:
   • Headline
   • CTA buttons
   • Adds a simple banner
4. A modified version of the page is returned

---

Features

• Works with real websites (including dynamic ones)
• Keeps original page structure intact
• Simple fallback logic if AI APIs are unavailable
• Generates a working personalized page instead of a mockup

---

Running locally

1. Clone the repository

git clone <repo>
cd AutoWebsite-Maker

---

2. Backend setup

cd backend
pip install -r requirements.txt
uvicorn main:app --reload

---

3. Frontend

The frontend is already built and stored in:

backend/static/

So no separate frontend setup is required.

---

Deployment

The app is deployed on Render using Docker.

To avoid build issues on low-memory environments:

• The frontend is built locally
• The build files are copied into backend/static/
• The backend serves these files directly

This keeps deployment simple and reliable.

---

Live Demo

https://autowebsite-maker-1.onrender.com

---

API Endpoints

• POST /analyze-ad
• POST /scrape-landing-page
• POST /generate-personalized-page
• GET /generated/{id}.html

---

Notes

• AI usage is optional — the system still works with fallback logic
• Some complex websites may not fully render as expected
• The project focuses more on the concept than full production readiness

---

License

MIT License