MyRealTrip's subsidiary AICX automates 75,000 CS cases per month

AICX is an AI Operations Center validated with MyRealTrip's operational data, starting as MRTCX in 2022 and expanding in 2024 into a customer experience and service operations company for external enterprises. Its core is a hybrid model combining its self-developed AICX Agent with human operators.

MyRealTrip is a platform with 2 trillion KRW in annual transaction volume and 6 million annual bookings, receiving 75,000 CS inquiries per month across phone and chat combined. Many inquiries—refunds, itinerary changes, local issues—involve complex context, making it difficult to handle them with a simple FAQ bot alone.

Automation began by layering an AI Agent on top of the existing counseling environment. A month's worth of inquiry logs were classified, and structured inquiries such as booking confirmation, voucher re-issuance, re-guiding product information, and itinerary change procedures were handed over first, while refunds, disputes, and safety issues entangled with local circumstances were left to humans.

• The chat AI chatbot AI Travel Mate Mari handles 80% of first responses, and resolves half of those on its own.
• The phone Agent takes 100% of first responses for non-urgent inquiries and resolves 57% on its own.
• Consultation context and booking information are passed to the agent's screen so the same question doesn't need to be repeated.
• The number of simple-inquiry agents fell by nearly half in six months, from 57 in July 2025, while customer satisfaction was similar between the AI chatbot at 4.61 and agents at 4.63.

Operational automation was also organized into two tracks. Email, messenger, internal tools, and partner channels were unified into a single queue, enabling 24-hour uninterrupted processing, and review, registration, and translation tasks that used to run on spreadsheets were moved to AI and rule-based automation.

Product development for new city categories was broken down into a 7-step pipeline, from city search to detail generation, with automation handling structured metadata processing while humans remained responsible for policy interpretation and quality judgment. The key was not the tools themselves but precisely drawing the boundary between humans and automation using operational data.

AICX is also applying this approach to external companies' AX (AI Transformation) and training programs.