🧠 DocMind AI — Advanced RAG Intelligence Platform

Built with Endee Vector Database | 20 AI-powered features | FastAPI + Vanilla JS

---

🚀 What It Does

DocMind AI is a full-stack RAG (Retrieval Augmented Generation) platform that transforms any PDF document into an interactive AI-powered knowledge base. Upload a PDF and instantly unlock 20 intelligent features.

---

✨ Features

Feature	Description
💬 AI Chat	Ask questions, get answers with source references
🚨 Hallucination Detector	Verifies every AI answer against document
⚡ Confidence Score	HIGH / MEDIUM / LOW rating per answer
📍 Source Highlighter	Shows exact page and passage used
💡 Smart Question Suggester	Auto-suggests 5 questions on upload
💬 Chat Memory	Remembers conversation context
🎓 Tutor Mode	Socratic teaching style
🌐 Web + Doc Hybrid	Combines document + live web search
🗺️ Mind Map	Visual topic structure
🕸️ Knowledge Graph	Entity relationship visualization
😊 Tone Analyzer	Sentiment, emotions, formality
🃏 Flashcard Generator	Auto study cards (exportable CSV)
📝 Quiz Generator	MCQ quiz with scoring
📅 Timeline Extractor	Dates and events visual timeline
✅ Action Items	Extracts tasks and to-dos
🥊 Document Debate	Compare two documents head-to-head
📧 Executive Email	One-click professional summary email
🔍 Facts vs Opinions	Color-coded classification
⚠️ Contradiction Finder	Finds conflicting statements
📚 Multi-Doc Search	Search across all documents simultaneously

---

🏗️ System Design

[User Browser]
     │
     ▼
[Frontend: HTML/CSS/JS]
     │  REST API calls
     ▼
[FastAPI Backend]
     │
     ├── /api/documents  → Upload, Ask, Multi-doc
     ├── /api/analysis   → MindMap, Tone, Debate, Email...
     ├── /api/quiz       → Flashcards, Quiz, Score
     └── /api/search     → Semantic, Hybrid, Evolution
     │
     ├── [Endee Vector DB] ← PDF chunks stored as embeddings
     ├── [OpenAI GPT-4o]   ← LLM for answers & analysis
     └── [SerpAPI]         ← Optional web search

---

🔧 How Endee Is Used

Endee serves as the core vector database for this platform:

1. PDF text is split into chunks (1000 chars, 200 overlap)
2. Each chunk is embedded using OpenAI text-embedding-3-small
3. Embeddings are stored in Endee vector collections
4. On every question, the query is embedded and Endee performs similarity search
5. Top-K most relevant chunks are retrieved and passed to GPT-4o

---

⚙️ Setup Instructions

1. Clone & Fork

# Star and fork https://github.com/endee-io/endee first!
git clone https://github.com/YOUR_USERNAME/endee.git
cd endee

2. Install Backend

cd advanced-rag-platform/backend
pip install -r requirements.txt

3. Configure Environment

cp .env.example .env
# Edit .env and add your keys:
# OPENAI_API_KEY=sk-...
# SERPAPI_KEY=...  (optional, for web search)

4. Run Backend

python main.py
# Server starts at http://localhost:8000

5. Open Frontend

Open browser → http://localhost:8000

---

📁 Project Structure

advanced-rag-platform/
├── backend/
│   ├── main.py                    # FastAPI app entry point
│   ├── config.py                  # Configuration & env vars
│   ├── requirements.txt
│   ├── .env.example
│   ├── routers/
│   │   ├── documents.py           # Upload, ask, multi-doc
│   │   ├── analysis.py            # All analysis features
│   │   ├── quiz.py                # Quiz & flashcards
│   │   └── search.py              # Semantic & hybrid search
│   ├── services/
│   │   ├── rag_service.py         # Core RAG pipeline
│   │   ├── hallucination_service.py
│   │   ├── analysis_service.py    # MindMap, KG, Tone...
│   │   ├── flashcard_service.py
│   │   └── web_search_service.py
│   └── utils/
│       ├── embeddings.py          # Endee vector store
│       └── pdf_processor.py       # PDF text extraction
└── frontend/
    ├── index.html                 # Single page app
    ├── styles.css                 # Dark theme UI
    └── app.js                     # All frontend logic

---

🛠️ Tech Stack

Layer	Technology
Vector DB	Endee (ChromaDB compatible)
LLM	OpenAI GPT-4o
Embeddings	OpenAI text-embedding-3-small
Backend	FastAPI + Python
Frontend	Vanilla JS + HTML/CSS
PDF Processing	pdfplumber + PyPDF2
Knowledge Graph	vis.js network
Web Search	SerpAPI (optional)

---

📸 Screenshots

Upload PDF → Ask questions → Get verified answers with source references, hallucination detection, and 20 more intelligent features.

---

👨‍💻 Author

Built for the Endee.io Internship Assessment
GitHub: Pradeepit2004

---

⭐ Don't forget to star endee-io/endee!