MegazoneCloud validates innovative potential of generative AI in financial sector

The use of generative artificial intelligence (AI) technology in the financial sector is accelerating. Following the Financial Services Commission's announcement of the 'Financial Sector Network Separation Improvement Roadmap' in August last year, the migration of financial data to the cloud has become possible, and financial companies are making successive attempts to improve efficiency and competitiveness through AI adoption..

Network separation regulations are a system that strengthens security by physically separating the internal and external networks of financial companies..While it has contributed to protecting financial systems, it has served as an obstacle to the rapid adoption of innovative technologies such as cloud and AI. Financial authorities have also introduced phased improvement measures to ease network separation so that financial companies can smoothly utilize generative AI in cloud-based internet environments and pursue digital transformation (DX)..

Accordingly, the financial sector is actively moving to adopt generative AI in internal operations. DX is taking place in various fields such as business automation, data management, and customer service, and digital innovation specialists are playing a key role in realizing this..

However, for financial companies to adopt AI, they must carefully review security regulations and practical applicability. MegazoneCloud, which has built expertise and technological capabilities in the financial IT field, is supporting financial companies in applying AI by filing for financial regulatory sandboxes and conducting PoC (Proof of Concept) to explore ways to maximize efficiency through AI application..

In particular, MegazoneCloud applies its self-developed generative AI service 'GenAI360' to enable financial companies to maximize work efficiency and productivity while complying with strict security regulations. GenAI360 is MegazoneCloud's all-in-one AI service that helps build generative AI systems tailored to the characteristics of enterprises and industries, providing integrated support from consulting to development and operation. In October last year, the company launched 'GenAI360 v2,' which added a 'GenAI360 Platform' that increases the reliability and optimization of generative AI without programming knowledge..

Generally, building generative AI requires long development periods and high costs. However, rather than perfectly building vast amounts of data in advance, MegazoneCloud applies an approach that first utilizes core data to build a RAG (Retrieval-Augmented Generation) pipeline and then gradually expands it..

Additionally, through consultation with customers, the company sets the direction for AI adoption, checks the data readiness status, and then proceeds with generative AI construction projects tailored to the required business areas. A test environment can be built in as little as 5 days, and projects can be completed within 4 weeks..

MegazoneCloud has conducted AI projects across various industries using GenAI360. Last year, using GenAI360, the company conducted Happy Campus knowledge product generation service, Hana Tour AI chatbot system, hy (formerly Korea Yakult) in-house mall search system improvement, and Hansol Paper sales log system advancement projects, executing numerous generative AI projects. More than half of these have led to actual business value creation beyond PoC..

MG Saemaul Geumgo Central Association, Considering Adoption of Employee AI Chatbot to Enhance Work Efficiency

MG Saemaul Geumgo Central Association recently collaborated with MegazoneCloud to conduct a generative AI-based employee work knowledge Q&A AI chatbot project. Until now, MG Saemaul Geumgo employees had to confirm unknown work knowledge by calling central association staff members operating by region or by searching for work knowledge through internal work portals.

MegazoneCloud utilized the 'GenAI360' platform in developing the AI chatbot for MG Saemaul Geumgo. MegazoneCloud analyzed MG Saemaul Geumgo's existing data usage and management processes, identified problems in the process of employees searching and utilizing data, and then proceeded to develop an employee AI chatbot using generative AI technology to address these issues..

In this process, using Vector Database (Vector Database), financial documents, legal materials, and product guides were optimized and converted to enable AI learning. Vector DB converts vast amounts of text and image data into vector form that AI can easily understand and stores it, enabling fast search of semantically similar information. For example, just as humans naturally think of terms like 'repayment,' 'debt,' and 'delinquency' when they see the word 'loan,' Vector DB also converts data into numbers, stores it, and quickly finds related information..

To prevent errors such as recommending financial products that do not actually exist, MegazoneCloud also applied RAG technology. Rather than the Large Language Model (LLM) simply responding based on pre-trained data, it searches for related information in the internal database using Vector DB and generates more accurate responses by reflecting this information..

Testing was conducted in the actual usage environment to improve and verify the accuracy and effectiveness of the chatbot. When a customer inquires about a specific product, employees can immediately check related terms, product descriptions, and other necessary information through the AI chatbot. When employees have questions about internal regulations or related laws, they can receive reliable official documents directly from the chatbot..
An official from MG Saemaul Geumgo Central Association stated, "We confirmed that generative AI can provide more accurate information based on MG Saemaul Geumgo's work-related document searches," and added, "Through the project with MegazoneCloud, we were able to confirm how much the adoption of generative AI can help improve employee work efficiency."
 
JB Woori Capital, Collaborating with MegazoneCloud on Generative AI-Based Report Automation

JB Woori Capital also embarked on collaboration with MegazoneCloud to review improving financial work efficiency through document writing automation. Based on the judgment that existing repetitive document work is causing productivity decline and resource waste, the company collaborated with MegazoneCloud to conduct verification of automatic report generation functionality using generative AI.

In this process, the company confirmed ▲automatic report generation including tables and images after analyzing original data ▲systematic organization of documents ▲improved readability by applying visual elements such as shading. The company also conducted technical review on ways to add automatic response functions for internal regulations, work guidelines, and welfare systems to enable employees to quickly confirm necessary information..

An official from JB Woori Capital's Digital Planning Team stated, "We confirmed that generative AI can lead to improvements in internal processes and enhanced customer service quality," and added, "Based on this achievement, we plan to undergo more rigorous verification steps and expand to more business areas."

James Kong, Center Director of MegazoneCloud's AI & Data Analysis Center, stated, "MegazoneCloud is continuously researching and developing financial company-customized AI solutions with enhanced security and data governance environments," and added, "Going forward, we will continue to support financial companies in realizing business innovation through AI technology and lead the financial industry's DX by exploring ways for generative AI to be safely and effectively utilized in the financial sector.".

Article Source: IT Chosun