Cognitive Buyer Intent Engine (CBIE)

A Production-Ready AI SalesIQ Bot for Zoho Cliqtrix

Key Features

• Intent Detection: Uses XGBoost to classify users as "Ready to Buy", "Researching", "Support Needed", etc.
• Sentiment Analysis: Real-time emotion detection (Positive, Negative, Neutral) using DistilBERT.
• Smart Recommendations: Hybrid recommender system suggesting products based on user behavior.
• Dynamic Personality: The bot adapts its tone (Friendly, Professional, Empathetic) based on user sentiment.
• Zoho Integration: Seamlessly integrates with SalesIQ via Deluge and invokeUrl.

Project Structure

cbie_system/
├── backend/                 # Python FastAPI Backend (The Brain)
│   ├── app/                 # API Logic
│   ├── ml_engine/           # Machine Learning Models & Training Scripts
│   └── requirements.txt     # Python Dependencies
├── zoho_deluge/             # Deluge Scripts (The Body)
│   ├── bot_handler.deluge   # Main Zobot Script
│   └── functions/           # Helper Functions
├── docs/                    # Documentation & Guides
└── dashboard/               # Analytics Dashboard Designs

Quick Start

1. Backend Setup

1. Navigate to backend/:

   cd backend

2. Install dependencies:

   pip install -r requirements.txt

3. Run the server:

   uvicorn app.main:app --reload

   The API will be available at http://localhost:8000.

2. Zoho SalesIQ Setup

1. Create a new Zobot in SalesIQ.
2. Choose Deluge as the platform.
3. Copy the content of zoho_deluge/bot_handler.deluge into the main handler.
4. Create a function named invoke_cbie_backend and paste the content from zoho_deluge/functions/invoke_backend.deluge.
5. Update the backend_url in the function to point to your deployed backend (e.g., on Render/Vercel).

Machine Learning

• Train Intent Model: python backend/ml_engine/training/train_intent.py
• Train Recommender: python backend/ml_engine/training/train_recommender.py

Dashboard

Check dashboard/ for analytics mockups and implementation details.

License

MIT License. Built for Zoho Cliqtrix Hackathon.