SmartDocs Assistant

A powerful RAG (Retrieval Augmented Generation) chatbot with conversational booking capabilities, built with FastAPI, Streamlit, and Google AI.

🚀 Features

📚 Document Q&A System

• Multi-format Support: Upload PDF and TXT documents
• Intelligent Chunking: Smart document segmentation with metadata preservation
• Vector Search: Semantic search using Google's text-embedding-004
• Citation Tracking: Source attribution with page numbers and snippets
• Conversational Memory: Context-aware follow-up questions

📅 Conversational Booking Agent

• Natural Language Processing: Book calls through natural conversation
• Flexible Information Collection: Collect name, phone, email, and date/time at any point
• Smart Date Parsing: Understand "tomorrow", "next Monday at 3pm", etc.
• Google Calendar Integration: Automatic calendar event creation
• Booking Persistence: Save and track all bookings

🎯 Key Capabilities

• Intent Detection: Automatically detect booking vs. document queries
• Basic Conversation: Handle greetings, gratitude, and casual chat
• Real-time Chat Interface: Modern Streamlit UI with conversation history
• Document Management: Upload, clear, and track document status
• Debug Tools: Built-in testing and verification tools

🏗️ Architecture

SmartDocs Assistant -monorepo/
├── apps/
│   ├── api/                    # FastAPI backend
│   │   ├── main.py            # API entry point
│   │   ├── routers/           # API endpoints
│   │   └── models/            # Request/response models
│   └── frontend/              # Streamlit frontend
│       └── streamlit_app.py   # Chat interface
├── packages/
│   ├── rag/                   # RAG system
│   │   ├── ingestion/         # Document processing
│   │   ├── retrieval/         # Vector search
│   │   └── chains/            # QA chains
│   ├── agents/                # Booking agent
│   │   ├── booking_agent.py   # Main agent logic
│   │   ├── state.py           # Conversation state
│   │   └── tools/             # Agent tools
│   └── shared/                # Shared utilities
├── data/                      # Document storage
├── tests/                     # Test suite
└── docs/                      # Documentation

🛠️ Technology Stack

Backend

• FastAPI: High-performance API framework
• Qdrant: Vector database for embeddings
• Google AI: Gemini-1.5-flash for LLM, text-embedding-004 for embeddings
• LangChain: LLM orchestration and prompt management

Frontend

• Streamlit: Interactive web interface
• Chat Interface: Real-time conversation with message history

Data Processing

• pypdf: PDF parsing and text extraction
• RecursiveCharacterTextSplitter: Intelligent text chunking
• UUID5: Stable chunk identification

Booking System

• Google Calendar API: Real calendar event creation
• Pydantic: Data validation and serialization
• JSON Storage: Local booking persistence

📋 Prerequisites

• Python 3.12+
• Google AI API key
• Qdrant vector database
• Google Calendar API (for booking feature)

🚀 Quick Start

1. Clone and Setup

git clone <repository-url>
cd SmartDocs Assistant -monorepo
uv sync

2. Environment Configuration

Create .env file:

# Google AI
GOOGLE_API_KEY=your_google_api_key
GOOGLE_LLM_MODEL=gemini-1.5-flash
GOOGLE_EMBEDDING_MODEL=text-embedding-004

# Qdrant
QDRANT_URL=http://localhost:6333
QDRANT_COLLECTION=smartdocs

# Google Calendar (for booking feature)
GOOGLE_CALENDAR_CREDENTIALS_FILE=path/to/credentials.json
GOOGLE_CALENDAR_TOKEN_FILE=path/to/token.json

3. Start Services

# Start Qdrant (Docker)
docker run -p 6333:6333 qdrant/qdrant

# Start FastAPI backend
uv run python apps/api/main.py

# Start Streamlit frontend
uv run streamlit run apps/frontend/streamlit_app.py

4. Access the Application

• Frontend: http://localhost:8501
• API Docs: http://localhost:8000/docs

📖 Usage Guide

Document Q&A

1. Upload Documents

   • Navigate to the upload section
   • Select PDF or TXT files
   • Click "Upload" to ingest documents

2. Ask Questions

   • Use the chat interface to ask questions
   • Get answers with source citations
   • View confidence scores and source snippets

Booking Calls

1. Start Booking

   • Say "I'd like to book a call" or similar
   • Provide information naturally: "My name is John, email is john@example.com"

2. Flexible Information Collection

   • Provide details in any order
   • Use natural language: "tomorrow at 3pm"
   • Agent will collect missing information

3. Confirmation

   • Review booking details
   • Calendar event is automatically created
   • Receive booking ID and meeting link

Document Management

• Upload: Add new documents to the knowledge base
• Clear: Remove all documents from the system
• Status: View current document count in sidebar

🔧 API Endpoints

Document Management

• POST /upload - Upload and ingest documents
• DELETE /upload/clear - Clear all documents
• GET /upload/status - Get document status

Chat & Search

• POST /chat - Main chat endpoint (RAG + Booking)
• GET /search - Raw vector search
• GET /chat/bookings - Get all bookings

Response Format

{
  "answer": "Response text",
  "confidence": 0.95,
  "citations": [
    {
      "page": 1,
      "snippet": "Relevant text snippet...",
      "filename": "document.pdf"
    }
  ],
  "intent": "rag|booking|booking_complete",
  "booking_state": {...}
}

📊 Performance

Document Processing

• Chunking: ~1000 characters per chunk with overlap
• Embedding: Google's text-embedding-004 (768 dimensions)
• Search: Top-K retrieval with configurable K (default: 6)

Response Times

• RAG Queries: 2-5 seconds (embedding + retrieval + generation)
• Booking Agent: 1-3 seconds (intent detection + state management)
• Document Upload: 10-30 seconds (parsing + chunking + embedding)

🚧 Development

Project Structure

├── apps/           # Application layer
├── packages/       # Core packages
├── data/           # Data storage
├── docs/           # Documentation
└── scripts/        # Utility scripts

Contributing

1. Fork the repository
2. Create a feature branch
3. Make your changes
4. Add tests
5. Submit a pull request