Problem

As a freelancer, I worked with a client, a large financial institution, which was struggling to keep up with the high volume of customer inquiries they were receiving. Their customer service team was overwhelmed and unable to respond to all the requests in a timely manner. This led to frustrated customers and a decrease in overall satisfaction with the company.

I was tasked to to find a solution that would allow the client to handle a larger volume of inquiries while still providing excellent customer service.

Solution

I proposed developing and deploying a custom chatbot using the LLAMA model. The chatbot would be able to extract relevant documents from the client's knowledge base and use them to answer customer inquiries quickly and accurately.

To develop the chatbot, we used langchain and huggingface for natural language processing, llama index for document retrieval, pinecone database for storage, and flask for deployment. We also trained the chatbot using RAG (Retrieval-Augmented Generation) which allowed it to provide more detailed responses than traditional chatbots.

Impact

Reduced customer inquiry response time by 10%

Customer Satisfaction with a RAG Chatbot