PDF Chatbot

Upload Analyze Ask Understand PDFs with AI

PDF Chatbot – Upload, Analyze & Chat with PDFs

AI-powered web app to upload PDFs, analyze them with AI, and ask questions about their content.
No in-page PDF viewer — just clean, fast, and interactive analysis.

Live Demo: https://pdf-chatbot-chi.vercel.app/

---

Table of Contents

• Quick Start
• Why PDF Chatbot?
• Demo & Screenshots
• System Architecture
• Features at a Glance
• Tech Stack
• Installation
• Usage
• Project Structure
• Security & Privacy
• Roadmap
• Troubleshooting
• Contributing
• Acknowledgements
• License
• Author

---

Quick Start

git clone <your-repo-url>
cd PDF_CHATBOT
docker build -t pdf-chatbot-backend ./backend
docker run -e PORT=8000 -p 8000:8000 pdf-chatbot-backend

Open the frontend at http://localhost:8080 and start chatting with PDFs

( Back to top)

---

Why PDF Chatbot?

PDFs contain valuable information, but extracting insights from long documents is time-consuming and frustrating.

PDF Chatbot solves this by allowing you to:

• Upload multiple PDFs
• Define a persona and goal
• Instantly receive structured insights, summaries, and answers

No scrolling. No guessing. Just answers.

---

Demo & Screenshots

🎥 Demo & Screenshots

🏠 Home Dashboard

Clean landing dashboard with quick actions for PDF analysis

---

📂 Upload PDFs

Drag-and-drop PDF upload with multi-file support

---

⚙️ Analysis Settings (Persona & Task)

Define persona and job-to-be-done for focused analysis

---

💬 Ask Anything Interface

Ask natural language questions across uploaded PDFs

---

🧠 AI-Powered Insights & Results

Structured explanations, summaries, and extracted sections

( Back to top)

---

System Architecture

Flow Overview

1. User uploads PDFs via frontend
2. Backend extracts text, structure & tables
3. Semantic ranking identifies relevant sections
4. Persona & task guide AI output
5. Results returned as structured JSON

( Back to top)

---

Features at a Glance

• Multi-PDF Upload & Analysis
• Persona-driven AI understanding
• Semantic search & ranking
• Structured summaries & explanations
• Fast, no embedded PDF viewer
• Fully Dockerized backend
• Light & Dark mode UI

( Back to top)

---

Tech Stack

Frontend

• HTML, CSS, Vanilla JavaScript (fully custom, no frameworks)
• Font Awesome (CDN) for icons
• Custom animations & responsive design with pure CSS
• Fetch API + XMLHttpRequest for backend communication and progress tracking
• Static site — runs without build tools or bundlers

Backend

• Python 3.9+, FastAPI 0.110.0, Uvicorn 0.29.0
• PyMuPDF, pdfplumber, Pillow, pytesseract
• sentence-transformers, torch, scikit-learn, nltk, numpy<2.0
• CORS enabled
• Dockerized for deployment

Deployment

• Frontend Vercel
• Backend Render

---

Installation

1. Clone the repository

   git clone <your-repo-url>
   cd PDF_CHATBOT

2. Build the backend Docker image:

   docker build -t pdf-chatbot-backend ./backend

3. Run the backend:

   docker run -e PORT=8000 -p 8000:8000 pdf-chatbot-backend

4. Serve the frontend:
   • Use VSCode Live Server, Python http.server, or any static server:

     cd frontend
     python -m http.server 8080
     # or use Live Server extension in VSCode

   • Open http://localhost:8080

---

Usage

1. Click Upload PDFs drag & drop or select files
2. Click Done to upload
3. Click Analyze Collection, enter a persona and job/task, then click Start Analysis
4. Or click Quick Summary for fast document takeaways
5. Use Ask Anything to query explanations

---

Project Structure

PDF_CHATBOT/
 backend/
    main.py                  # FastAPI app
    analyze_collections.py   # Semantic analysis
    heading_extractor.py     # PDF structure extraction
    summary.py               # Summary generation
    explain.py               # Explanations
    setup_offline_assets.py  # Offline cache/setup
    requirements.txt         # Python dependencies
    Dockerfile               # Container build file
 frontend/
    index.html
    script.js
    style.css
    ...
 README.md
 ...

---

Security & Privacy

• No in-browser PDF embedding
• Files processed securely on backend
• No chat history stored
• Minimal local storage usage

( Back to top)

---

Troubleshooting

• If Analyze button is disabled: select at least one PDF and fill both persona & job fields
• If analysis fails: check backend logs, CORS, or API URL in script.js
• For Docker issues: ensure ports are mapped and backend is running

---

Roadmap

• RAG-based long-context QA
• Multi-language support
• Persona presets
• Confidence scoring & citations
• Offline mode

( Back to top)

---

Contributing

Contributions are welcome!
If you'd like to improve this project, please follow these steps:

1. Fork the repository
2. Create a feature branch (git checkout -b feature-name)
3. Commit your changes (git commit -m 'Add new feature')
4. Push to your branch (git push origin feature-name)
5. Open a Pull Request

---

Acknowledgements

This project wouldnt be possible without these amazing tools and libraries:

• FastAPI – Backend framework
• Sentence Transformers – Embeddings and NLP
• PyMuPDF – PDF parsing
• pdfplumber – Text extraction
• pytesseract – OCR
• Tailwind CSS & Bootstrap – Styling
• Font Awesome – Icons

---

License

MIT

Author

• Developed by Gudiwada Sruthi