📑 Agentic-RAG

A Streamlit-based AI assistant that helps users search and query PDF documents using Retrieval-Augmented Generation (RAG). It uses Qdrant as a vector store, SentenceTransformers for embeddings, and the Agno agentic framework to manage reasoning and tool orchestration with Azure OpenAI models.

---

🚀 Features

• Upload and index any PDF document
• Ask natural language questions about the document
• Retrieves relevant paragraphs using vector search (Qdrant)
• Uses Agno agents to decompose complex queries and route them to reasoning tools
• Generates accurate answers with confidence scores
• Built-in retry logic and robust fallback handling

---

🧱 Tech Stack

• Frontend: Streamlit
• Backend: Python
• Agent Framework: Agno
• Vector DB: Qdrant
• Embeddings: sentence-transformers/all-MiniLM-L6-v2
• LLM: Azure OpenAI GPT (configurable)

---

📦 Setup Instructions

1. 🔧 Clone the Repository

# Clone the repository
git clone https://github.com/your-username/document-assistant.git
cd document-assistant

# Create a virtual environment and activate it
python -m venv venv
source venv/bin/activate     # For Windows: venv\Scripts\activate

# Install dependencies
pip install -r requirements.txt

# Start Qdrant (requires Docker)
docker run -p 6333:6333 -v $(pwd)/qdrant_storage:/qdrant/storage qdrant/qdrant

# Run the Streamlit app
streamlit run src/app.py