chatbot_README.md

metadata

title: Budget Proposals Chatbot API
emoji: 🤖
colorFrom: green
colorTo: blue
sdk: gradio
sdk_version: 5.44.0
app_file: app.py
pinned: false

Budget Proposals Chatbot API

An intelligent chatbot API for budget proposals using LangChain, Gemini AI, and RAG (Retrieval-Augmented Generation) with conversation memory.

Features

• RAG-powered responses using semantic search of budget proposals
• Conversation memory with session management
• Gemini AI integration for natural language processing
• Source citation for all responses
• Session management with chat history
• Suggested questions for easy interaction

API Endpoints

Chat

• POST /api/chat - Chat with memory and RAG
• POST /api/chat/clear - Clear chat memory

Session Management

• GET /api/chat/sessions - List active sessions
• GET /api/chat/history/<session_id> - Get chat history

Utility

• GET /api/chat/health - Health check
• GET /api/chat/suggestions - Get suggested questions
• GET /api/chat/available-pdfs - Get available PDF files

Environment Variables

Set these in your Hugging Face Spaces secrets:

• GEMINI_API_KEY - Your Google Gemini API key (required)
• PINECONE_API_KEY - Your Pinecone API key (required)

Usage Examples

Chat with the Bot

curl -X POST https://your-space-url.hf.space/api/chat \
  -H "Content-Type: application/json" \
  -d '{"message": "What are the maternity leave benefits?", "session_id": "user123"}'

Clear Chat Memory

curl -X POST https://your-space-url.hf.space/api/chat/clear \
  -H "Content-Type: application/json" \
  -d '{"session_id": "user123"}'

Get Chat History

curl https://your-space-url.hf.space/api/chat/history/user123

Response Format

{
  "response": "Based on the budget proposals database, the maternity leave benefits proposal...",
  "confidence": "high",
  "session_id": "user123",
  "conversation_length": 5,
  "memory_used": true,
  "rag_used": true,
  "sources": ["MLB.pdf"],
  "user_message": "What are the maternity leave benefits?"
}

Integration

This chatbot integrates with the semantic search API to provide accurate, context-aware responses about budget proposals. It maintains conversation context and can reference previous interactions.

Deployment

This API is deployed on Hugging Face Spaces and automatically serves on port 7860.