AI-Powered Intelligent Brochure Generator

📖 AI-Powered Intelligent Brochure Generator

An AI-powered web application that transforms a company website into a professional brochure using web scraping, Large Language Models (LLMs), Hugging Face, and automated PDF generation.

This project was built as a learning project while following the AI Engineering Core Track by Ed Donner, with a focus on understanding how different AI engineering components can be connected into an end-to-end application.

🚀 Project Overview

The AI Brochure Generator allows a user to enter a company website URL and automatically generate brochure content from the information available on that website.

Workflow

Company Website
       ↓
Website Scraping
       ↓
Relevant Information Extraction
       ↓
Groq LLM
       ↓
Structured Brochure Content
       ↓
AI Image Generation
       ↓
Professional PDF Brochure
       ↓
Download

The goal is to make the entire process automatic so that users don't need to manually research a company and design a brochure.

✨ Features

• 🌐 Scrape information from company websites
• 🔎 Extract useful website content
• 🧠 Generate brochure content using an LLM
• 📦 Generate structured JSON output
• 🏢 Generate company overview
• 🛠️ Extract products and services
• ⭐ Extract key features
• 🎯 Generate company mission
• ℹ️ Extract important company information
• 🎨 Generate AI-powered brochure images
• 📄 Automatically generate a PDF brochure
• 🖥️ Interactive Streamlit interface
• 🔐 Environment-based API key management
• ☁️ Designed for deployment on Streamlit Cloud

🧠 Technologies Used

🤖 AI Pipeline

1. Website Scraping

The user provides a website URL through the Streamlit interface.

The application attempts to collect useful information such as:

• Website title
• Headings
• Paragraphs
• Lists
• Important links
• Relevant pages

For websites where normal HTTP scraping is insufficient, browser automation can be used as a fallback.

2. LLM Content Generation

The extracted website information is sent to a Groq-hosted LLM.

The LLM converts the raw website information into structured brochure content.

Example structure:

{
    "company_overview": "...",
    "products_services": [
        "...",
        "..."
    ],
    "key_features": [
        "...",
        "..."
    ],
    "mission": "...",
    "important_information": "..."
}

The system is instructed to use information supported by the scraped website and avoid unnecessarily inventing information.

3. AI Image Generation

The generated brochure information is converted into an image-generation prompt.

The prompt is then sent to a Hugging Face image-generation model.

The generated image is used as part of the brochure design.

4. PDF Generation

Finally, the structured brochure content and generated image are combined into a professional PDF.

The user can download the generated brochure directly from the Streamlit application.

🖥️ User Interface

The application provides a simple workflow:

Enter Website URL
       ↓
Click "Generate Brochure"
       ↓
Website Scraping
       ↓
AI Content Generation
       ↓
AI Image Generation
       ↓
PDF Generation
       ↓
Download Brochure

The interface also displays the progress of each stage so the user can understand what the application is currently doing.

📂 Project Structure

ai-brochure-generator/
│
├── app.py
│
├── main.py
│
├── llm.py
│
├── image_generator.py
│
├── pdf_generator.py
│
├── requirements.txt
│
├── .env
│
├── .gitignore
│
└── README.md

app.py

Responsible for the Streamlit user interface and connecting all parts of the application.

main.py

Acts as the main pipeline that connects scraping, LLM processing, image generation, and PDF generation.

llm.py

Handles communication with the Groq LLM and generates structured brochure content and image-generation prompts.

image_generator.py

Handles AI image generation through Hugging Face.

pdf_generator.py

Creates the final brochure PDF.

🔑 Environment Variables

Create a .env file in the project root:

GROQ_API_KEY=your_groq_api_key
HF_TOKEN=your_huggingface_token

Never commit your .env file to GitHub.

Add this to .gitignore:

.env
.venv/
__pycache__/
generated/

⚙️ Installation

1. Clone the repository

git clone https://github.com/Sakib-Hasan3/Agentic-AI-project.git

2. Go to the project directory

cd Agentic-AI-project

3. Create a virtual environment

python -m venv .venv

4. Activate the virtual environment

Linux/macOS:

source .venv/bin/activate

Windows:

.venv\Scripts\activate

5. Install dependencies

pip install -r requirements.txt

6. Configure API keys

Create .env:

GROQ_API_KEY=your_groq_api_key
HF_TOKEN=your_huggingface_token

7. Run the application

streamlit run app.py

The application will then be available through the local Streamlit URL.

📓 Kaggle Notebook

The initial AI/LLM experimentation and development for this project was performed in Kaggle.

👉 Kaggle Notebook: https://www.kaggle.com/code/mohammedsakibhasan3/ai-voucher

🌐 Live Demo

The project is designed to be deployed as a Streamlit application.

Live Demo: Add your deployed Streamlit Cloud URL here.

🧪 Learning Objectives

This project helped me practice several important AI Engineering concepts:

• Web scraping and data extraction
• Working with unstructured web data
• Prompt engineering
• Structured LLM outputs
• JSON parsing and validation
• API integration
• Hugging Face inference
• AI image generation
• PDF generation
• Streamlit application development
• Environment variable management
• Git/GitHub workflow
• Deploying AI applications

🎓 Learning Context

This project is part of my hands-on learning journey through the AI Engineering Core Track by Ed Donner.

Rather than only studying the concepts theoretically, I am building projects to understand how different AI components work together in real applications.

This is Project 1 of an 8-project learning track that I am working through to strengthen my practical AI Engineering skills.

🔮 Future Improvements

Planned improvements include:

• Better multi-page website crawling
• More reliable extraction from JavaScript-heavy websites
• Improved company information filtering
• Better brochure layouts
• Multiple PDF design templates
• Multiple AI image-generation options
• Improved image generation speed
• Better handling of websites with anti-bot protection
• More robust error handling
• Additional LLM providers
• Cloud deployment optimization

⚠️ Disclaimer

This is an educational and experimental project created for learning purposes.

The generated brochure content depends on the information available from the provided website and the capabilities of the underlying AI models.

👨‍💻 Author

Mohammed Sakib Hasan

Computer Science & Engineering Student Patuakhali Science and Technology University

GitHub: https://github.com/Sakib-Hasan3

🙏 Acknowledgment

Special thanks to Ed Donner for the practical and project-focused AI Engineering Core Track, which inspired this hands-on learning journey.