Sales Chatbot For An Insurance Company

Project description

The goal of the project was to create a Facebook Messenger chatbot Q&A system that would help them increase sales of insurance policies. Our chatbot used several Natural Language Processing algorithms and historical data to generate the most thoughtful answers. Machine learning models were trained using labeled data to detect the intent of a question. Upon the user’s request, the bot was able to transfer the user to a live support person to continue the dialogue.

The chatbot enabled clients to purchase insurance policies in less than 2 minutes. Was implemented an administrator panel for the chatbot where they were allowed to.   

• Manage some predefined questions and answers.
• Build a chat flow state machine for customization.
• Automated model training and deployment.
• Statistical reports.

Technologies:

• FastText, scikit-learn (For intent detection)
• Python, Waitress (Python WSGI server), MongoDB
• React.js for web UI