Project Overview
Project Summary :
nAvI is a service designed to address the issue of 'mis-selling of financial products' to the elderly. To solve this problem, the service features include providing basic information about products through search, extracting misguidance and correcting facts during conversations with bankers, and offering quizzes about the products being considered for enrollment. For easy explanations and the misguidance extraction feature, Google's generative AI, GEMINI, has been used, and image-generating AI has also been utilized.
Identifying the Challenge
The Social Problem:
Our problem is ‘Misselling to the elderly’ Mis-selling is The omission of crucial details or the misrepresentation and exaggeration of financial products by a financial company, which leads to consumer misunderstanding during the sales process. The reason why the problem is serious is that there is a large amount of damage of 7 trillion won, and it is a problem that is repeated continuously for 20 years.
Innovation and Uniqueness
Why Our Project Stands Out:
The most important part of our project is how to describe financial products easily. For this purpose, we put the most effort into utilizing generative ai and deriving accurate and practical answers to generative ai. For this purpose, we learned pdf and learned various reference data to increase accuracy.
Insights and Development
Learning Journey:
I realized the importance of communication and a quick and accurate information sharing system in the process of team project.
In particular, as the entire process was carried out within a short period of time, it had to be shared and applied to development as soon as the contents of the plan were revised.However, we overcame that difficulity through the way using a shared document service in which we organize everything about project.
Development Process:
We developed the Android app front with a flutter framework that provides a variety of libraries of app development essential functions to make apps easy and fast in a short period of time. The backend was developed with Java and Sprinsg Boot framework, using MySQL for the main DB. The DB was run with GCP Cloud SQL, and the Spring Boot application was built as a Docker image, pushed to GCP Artifact Registry, and deployed via Compute Engine.
