INTRODUCTION
LG U+, a key subsidiary of LG Corporation and a leader in mobile, internet, and AI transformation, operates one of the largest customer service centers in Korea, with 4,000 customer service agents handling 3.5 million calls per month.
In recent years, LG U+ has expanded its AI services, making significant strides with its innovative agent support solution: Agent Assist.
Built from scratch on MongoDB Atlas, the solution is redefining the role of AI in LG U+’s AI Contact Center. Agent Assist has helped improve resource efficiency by 30 percent, enhancing LG U+’s overall operations and ultimately elevating the customer experience.
THEIR CHALLENGE
Delays and challenging data retrieval were causing inefficiencies
LG U+’s commitment to digital transformation underpins a company-wide goal: to efficiently manage the increasing number of customer interactions and improve the quality of consultations. This objective requires the support of a robust, scalable database able to manage complex and unstructured data.
“Traditional call centers often struggle with inefficiencies, from slow response times to difficulty retrieving relevant information,” said Minkyu Ha, Senior Software Engineering Manager at LG U+. “Agent Assist addresses these challenges with real-time transcription, automatic categorization of customer inquiries, and intelligent contextual summaries, ensuring smoother interactions between agents and customers.”
The company determined that the complex and inflexible schema of a relational database didn’t fit its needs . Particularly, LG U+ needed to run real-time AI features during consultations, store the results, and run multiple tests for service improvements, which the relational model didn’t accommodate.
“We knew we wanted to develop our new Agent Assist service on a document database. Furthermore, we needed a solution that would [enable] us to efficiently manage all the vector data generated during consultations and ensure smooth operations,” said Ha. “That’s why we chose MongoDB Atlas, the industry leader in this area.”
THE SOLUTION
MongoDB Atlas scales to seamlessly handle high call volumes
Within four months of implementing MongoDB Atlas, LG U+ was able to develop and launch the Agent Assist service.
“Although there were learning curves, especially for developers unfamiliar with MongoDB, its intuitive document model facilitated a smooth migration, enabling us to accelerate service development and testing,” said Ha.
With Agent Assist built on MongoDB Atlas, LG U+ is able to unleash the power of its AI solution. Agents can now access information faster and deliver more accurate responses with minimal delays.
To enhance advanced AI capabilities and workflows, LG U+’s AI engineering team also adopted MongoDB Atlas Vector Search early in development. This meant they could fully exploit the AI capabilities they were developing, such as identifying customer intent in real time and suggesting answer guidelines to agents.
Another major improvement is MongoDB’s ability to consolidate vector and operational data in a single system. “Managing both vector and operational data in MongoDB opened a new world to our team. By migrating operational data from PostgreSQL to MongoDB, we eliminated redundant processes and streamlined data queries,” said Ha.
THE OUTCOME
AI unleashed 30 percent higher efficiency and a better customer experience
Since launching Agent Assist with MongoDB, LG U+ has seen notable improvements.
“With MongoDB Atlas as the backbone supporting our AI features, the service has enhanced efficiency by reducing processing time by 7 percent on average per call.”
Furthermore, “Using MongoDB’s flexible document model to manage both vector and operational data has improved resource efficiency by 30 percent,” Ha added. This has enhanced speed and performance in LG U+’s new service development.
“MongoDB’s auto-scaling capability ensures seamless performance, even during peak morning hours, when call volumes are at their highest,” said Ha.
Beyond performance gains, MongoDB’s monitoring and real-time alert system has empowered the AI engineering team to quickly identify and resolve inefficiencies, like slow queries. This proactive approach to database management has strengthened system reliability and improved query latency, allowing them to see sub-second latencies for large amounts of $vectorSearch queries at scale. These capabilities also offer a better experience for both agents and customers.
Initially launched with 1,200 agents, the Agent Assist service is set to expand to 2,300 agents in the first half of 2025. This service currently serves over 1M user queries per week, and is projected to expand with the increase in the number of agents. In addition to refining Agent Assist, LG U+ is exploring improvements to its callbots and chatbots, using MongoDB’s flexible architecture to further enhance agent efficiency.
With its strong relationship with MongoDB, LG U+ is poised to advance intelligent customer interactions through its AI Contact Center and expand these services into the business-to-business market. This will help more businesses gain access to AI-powered customer service solutions that can transform customer engagement.
.svg)