# Automating Customer Service using LangChain: Building custom open-source GPT Chatbot for organizations

**Authors:** Keivalya Pandya, Mehfuza Holia  
**arXiv:** [2310.05421](https://arxiv.org/abs/2310.05421)  
**Date:** 2023-10-09  
**Summary page:** https://www.summarizepaper.com/en/paper/2310.05421v1/

## Key points

- Introduction of LangChain, a custom Large Language Model (LLM) for automating customer service
- Obsolescence of traditional customer support techniques like FAQs
- Proposal for responsive, context-aware, and personalized customer interactions
- Fusion of open-source methodologies, web scraping, fine-tuning, and seamless integration of LangChain into customer service platforms
- Demonstration of the framework "Sahaay" using Birla Vishvakarma Mahavidyalaya (BVM) Engineering website as an example context for the chatbot
- Literature survey discussing challenges and advancements in deploying large neural models for information retrieval
- Evaluation of LLM performance with comparative analysis using metrics; Google's Flan T5 XXL model outperforms other competitive LLMs like Base and Small
- Applications extend to various industries with round-the-clock support in multiple languages and addressing a wide array of inquiries and issues
- Technology leveraged to enhance the customer-company relationship, improve retention rates, extract higher value from customers, and elevate brand image.

## Summary

This research paper introduces a groundbreaking approach to automating customer service using LangChain, a custom Large Language Model (LLM) tailored for organizations. The paper highlights the obsolescence of traditional customer support techniques such as Frequently Asked Questions (FAQs) and proposes a paradigm shift towards responsive, context-aware and personalized customer interactions. The innovation lies in the fusion of open-source methodologies, web scraping, fine-tuning and the seamless integration of LangChain into customer service platforms. The framework known as "Sahaay" is demonstrated using Birla Vishvakarma Mahavidyalaya (BVM) Engineering website as an example context for the chatbot. This allows users to retrieve real-time information and receive answers to their queries. The paper also includes a literature survey that discusses the challenges and advancements in deploying large neural models for information retrieval. The methodology section evaluates the performance of LLMs and presents comparative analysis using metrics. Results show that Google's Flan T5 XXL model outperforms other competitive LLMs like Base and Small. The applications of this research extend to various industries with customer service being elevated through efficient, personalized and responsive interactions powered by LLMs like LangChain. Automated customer service offers round-the-clock support in multiple languages and can effectively address a wide array of inquiries and issues. Overall, this research heralds a new era in customer service where technology is leveraged to enhance the customer-company relationship, improve retention rates, extract higher value from customers and elevate brand image.