README.md

---
title: AI Chatbot with Smart Routing
emoji: 🤖
colorFrom: blue
colorTo: purple
sdk: docker
pinned: false
license: mit
app_port: 7860
---

# 🤖 Financial AI Chatbot with Smart Routing & RAG

**A demo GenAI app that demonstrates smart routing using LangChain**

## 🚀 Try It Live
- **🎯 Live Demo**: [Financial AI Chatbot](https://huggingface.co/spaces/krinya/smart_routing_with_render_example) ← **Try it here!**
- **💻 Frontend Code**: [`app.py`](https://huggingface.co/spaces/krinya/smart_routing_with_render_example/tree/main) - Gradio interface code
- **🔗 Backend API**: [Deployed on Render](https://gen-ai-demo-rag-bot.onrender.com/docs)
- **📖 Backend API Code**: [GitHub Repository](https://github.com/krinya/gen_ai_demo_rag_bot/tree/main)

## 🎯 What This Demonstrates

This project shows **how to create a complete GenAI product**:

### 1. 🧠 Smart Routing with LangChain
Intelligently routes financial questions about **5 major companies** (Apple, Google, Amazon, Tesla, Intel):
- 🔍 **FAQ Route**: Quick facts (CEO names, founding dates)
- 📚 **RAG Route**: Financial data from 2024 annual reports (revenue, profits)  
- 🧠 **LLM Route**: General explanations and financial concepts

### 2. 📊 RAG Implementation
- **Vector Storage**: ChromaDB with processed financial documents (full annual reports)
- **Retrieval System**: Semantic search for relevant information
- **Smart Fallbacks**: Multiple sources with quality scoring

### 3. 🏗️ Production Architecture
- **Backend**: Python FastAPI with LangChain, deployed on Render
- **Frontend**: Gradio UI deployed on Hugging Face Spaces  
- **Separation**: Backend API + Frontend UI for scalability

## 🛠️ How This Shows GenAI Product Development

**Complete workflow: Backend → Deploy → Frontend**

1. **Write Backend** (Python + LangChain)
   - FastAPI with smart routing logic
   - RAG pipeline with vector storage
   - Deploy on Render cloud platform

2. **Create Frontend** (Gradio + Hugging Face)
   - Interactive chat interface
   - Real-time routing insights
   - Deploy on Hugging Face Spaces

3. **Connect & Scale**
   - Backend API serves multiple frontends
   - Docker containerization
   - Production-ready architecture

## 🔧 Tech Stack

- **AI**: OpenAI GPT-4o-mini + LangChain orchestration
- **Backend**: Python FastAPI deployed on Render
- **Frontend**: Gradio deployed on Hugging Face Spaces
- **Storage**: ChromaDB vector database
- **Data**: 2024 financial reports (Apple, Google, Amazon, Tesla, Intel)

## � Example Queries

Try these in the live demo:
- "Who is the CEO of Tesla?" → FAQ route
- "What was Apple's revenue in 2024?" → RAG route  
- "How do you calculate P/E ratio?" → LLM route

---

**🎯 Key Learning**: This demonstrates the complete GenAI development stack from data processing to production deployment!