1. Overview (Project Background)

Company J operated a contact center serving customers of savings banks nationwide, providing financial product guidance, complaint registration, and procedural inquiries.

However, the short cycle of financial policy changes and complex procedures increased the knowledge burden on counselors, and the lack of systematic inquiry type classification made it difficult to standardize response quality.

Additionally, as customer inquiry volume increased, the need for real-time consultation chat environments (Chat Counseling) grew, and the company pursued the development of an AI-based automated chatbot and consultation chat system with the goals of providing accurate financial information, automation, and operational efficiency.

2. Challenge (Problem Definition)

• Difficulty in agile updates due to frequent changes in financial policies and products

• Necessity for procedure-based branching when providing sensitive financial information

• Difficulty in ensuring quality consistency due to variations in counselor responses

• Absence of inquiry type classification system → Increased unnecessary waiting time before counselor connection

• Increased need for consultation chat adoption → Lack of integration with customer information and history

• Existence of special requirements such as financial compliance, security, and access control

3. Solution (Resolution Approach)

• Construction of Rule-based Chatbot Based on Financial Policies, Products, and Procedures
  We automated the provision of accurate financial information to resolve basic inquiries through the chatbot.

• Implementation of Intent Classification for Automatic Inquiry Categorization System
  We analyzed natural language utterances to automatically identify categories such as complaints, products, and procedures, and standardized workflows.

• Real-time Financial API Integration
  We applied API integration to provide real-time financial information such as product inquiries, eligibility verification, and application availability.

• Introduction of Consultation Chat (Chat Counseling) Functionality

  • Seamlessly connect customers from chatbot to consultation chat for inquiries that cannot be resolved by the chatbot

  • Counselors can view customer utterance history, category, and progress status in real-time

  • Implemented UX that allows counseling to continue without unnecessary repeated questions

• Automatic Context Transfer When Connecting to Counselor
  We configured the system to automatically display utterances, intentions, and necessary financial information collected by the chatbot on the counselor's screen.

• Provision of Maintenance-Friendly Management Tools
  We provided an environment where operators can directly update scenarios and responses to suit the characteristics of financial organizations where policy changes and FAQ updates are frequent.

• Compliance with Financial Security, Logging, and Authority Standards
  We designed the system to meet financial institution compliance requirements such as access control, audit logging, and customer data management.

4. Result (Achievements)

• Increased Automatic Processing Rate for Repetitive Financial Inquiries → Reduced Counselor Inflow
  Simple and standardized inquiries are resolved through the chatbot, significantly reducing the workload of counselors.

• Improved Customer Experience Through Response Standardization
  Consistency in response quality has been ensured through automatic guidance based on financial regulations.

• Improved Real-time Consultation Efficiency Through Chat Counseling Adoption
  Customer history, utterances, and inquiry types are automatically transferred to counselors, resulting in reduced waiting time and improved consultation speed.

• Improved Policy Change Reflection Speed → Reduced Operational Burden
  Through management tool-based rapid updates, we can now respond agilely to financial policy changes.

• Secured Insights Based on Financial CS Data
  We established an environment for continuous service improvement by analyzing complaint types, FAQ patterns, and inquiry flows.