Building text-to-text conversational AI system for insurance renewal reminders | |
| Author | Nirut Gammayeengoen |
| Call Number | AIT Thesis no.DSAI-24-09 |
| Subject(s) | Insurance--Data processing Natural language generation (Computer science) Artificial intelligence |
| Note | A thesis submitted in partial fulfillment of the requirements for the degree of Master of Engineering in Data Science and Artificial Intelligence |
| Publisher | Asian Institute of Technology |
| Abstract | Customer retention is crucial in the insurance industry, with renewal reminders playing a pivotal role in maintaining customer relationships. Traditional call centers often face challenges handling high volumes and complex interactions, leading to inefficiencies, frustrated customers, and missed renewal opportunities. This thesis proposes the development of a text-to-text conversational AI system designed to automate and enhance the process of insurance renewal reminders.Theresearch focuses onintegrating suitable Large Language Models (LLMs), Retrieval Augmented Generation (RAG), Knowledge Graphs, and multi-agent systems to improve the AI’s performance and user experience. Key research questions include identifying cost-effective and accurate LLMs for different system modules (Qualifier, Reminder, Apologizer, Questioner, Oracle, Mailer, Goodbyer), exploring the use of LLM multi agents to enhance system functionality, and determining which modules can benefit from Knowledge Graph integration. The objectives of this study encompass investigating existing approaches, designing and building the core components of the conversational AI system, and evaluating its perfor mance through human and non-human (LLM-as-a-Judge) evaluations. The evaluation metrics will cover various categories, such as customer inquiries about policy details, renewal methods, insurance coverage, and handling of non-insurance-related questions.By leveraging prompt engineering techniques, contextual prompts, instruction-based prompts, role-playing prompts, and few-shot learning, this thesis aims to develop a robust conversational AI system that provides accurate, relevant, and user-friendly responses. The expected results include improved response accuracy, reduced operational costs, enhanced customer satisfaction, and increased efficiency in handling insurance renewal reminders. This research contributes to the broader field of conversational AI and its application in the insurance industry, offering a scalable and efficient solution to enhance customer engagement and retention. |
| Year | 2024 |
| Type | Thesis |
| School | School of Engineering and Technology |
| Department | Department of Information and Communications Technologies (DICT) |
| Academic Program/FoS | Data Science and Artificial Intelligence (DSAI) |
| Chairperson(s) | Chaklam Silpasuwanchai; |
| Examination Committee(s) | Attaphongse Taparugssanagorn;Chantri Polprasert; |
| Scholarship Donor(s) | AIS Scholarships; |
| Degree | Thesis (M. Eng.) - Asian Institute of Technology, 2024 |