'AI Consultation Advisor' Technology Briefing Held
Over 1.17 Million Minutes Saved Per Month After Advisor Adoption
Based on 'Exaone,' the LLM from LG AI Research
AI to Provide Consultation Feedback... Planned for Introduction in Q3
LG Uplus is applying its self-developed artificial intelligence (AI) to its customer service centers, helping to save customers' consultation time. Through the 'AI Consultation Advisor,' which understands customers' questions and recommends tailored consultation content, the company has reduced both waiting and consultation times by an average of approximately 1.17 million minutes per month.
On the morning of May 27, LG Uplus held a technology briefing at its headquarters in Yongsan-gu, Seoul, to introduce its self-developed 'AI Consultation Advisor.' The AI Consultation Advisor is an AI agent applied to the AI Contact Center (AICC) used by consultants, assisting them throughout the entire process?from the moment a customer call is received to the post-consultation tasks that need to be handled.
LG IT·() 27 AI Consultation Advisor held at the LG Uplus building in Yongsan-gu, Seoul. Photo by Lee Myunghwan
AICC refers to a customer center that utilizes AI. According to market research firm Allied Market Research, the domestic AICC market is expected to grow at an average annual rate of 23.7%, reaching approximately KRW 454.6 billion by 2030.
Since the AI Consultation Advisor was introduced to the LG Uplus customer center in September last year, the average connection waiting time per customer call has decreased by 17 seconds, and the average call duration has been reduced by 30 seconds. As a result, the total consultation time has decreased by about 19%. Considering that an average of about 75,000 consultations are received per day, the amount of consultation time saved in a month totals approximately 1.17 million minutes.
LG Uplus plans to further reduce the time it takes to connect to a consultant. Namhee Seo, CV Manager at the LG Uplus CX Center, said, "As the time required to handle a single call decreases, the time it takes to connect to a consultant when requested has also been reduced. In the past, customers had to wait up to one minute for a connection, but now that time has been reduced to just seven seconds."
Seo also emphasized, "To provide a differentiated customer experience, we aim to minimize the waiting time to zero from the moment the customer presses the consultant connection button until the connection is made. We also plan to eliminate the time customers are kept waiting while consultants search for information during consultations."
Built on 'Exaone' by LG AI Research... Utilizing Agentic RAG and AI In The Loop
On the morning of the 27th, the 'AI Consultation Advisor' was demonstrated at the technology briefing held at the headquarters in Yongsan-gu, Seoul. Photo by Lee Myunghwan
The core technologies of the AI Consultation Advisor developed by LG Uplus are 'Agentic RAG (Retrieval-Augmented Generation)' and 'AI In The Loop.' Agentic RAG is a technology in which AI autonomously analyzes the consultation between the consultant and the customer, searches for necessary information from within the company, and generates answers on its own. 'Agentic' refers to autonomous judgment and action, while RAG (Retrieval-Augmented Generation) is a generative AI that retrieves information from external sources; combined, they form the term 'Agentic RAG.'
Agentic RAG is characterized by its ability to independently assess situations and autonomously generate answers. For example, when a customer asks, "I've heard about eSIM?what is it?" the system not only provides an explanation of eSIM but also offers information about the subscription process and points to note. The process also includes a review to check whether the answer matches the intent of the question, further improving accuracy. According to LG Uplus's own analysis, the accuracy rate of answers provided by Agentic RAG reached 90%.
AI In The Loop is an AI technology designed to improve the accuracy of categorizing post-consultation conversations by topic. Previously, it was necessary to manually check and assess the accuracy of summaries and classifications made by AI, which took time. With AI In The Loop, another AI engine verifies the accuracy of the classifications, and if errors are detected, it learns and corrects itself.
LG Uplus reduced customers' consultation time by an average of approximately 1.17 million minutes per month by utilizing its self-developed AI consultation advisor. The photo is a screenshot of the actual operation screen of the AI consultation advisor. Provided by LG Uplus. Photo by LG Uplus
By utilizing AI throughout the entire post-consultation process, both classification accuracy and speed have improved. According to analysis by the LG Uplus customer center, it took approximately 5,760 minutes for a person to classify 2,000 consultations, whereas using AI In The Loop, 3,000 consultations could be processed in just 40 minutes.
LG Uplus also plans to develop and introduce 'AI Auto QA,' which allows AI to evaluate consultation content and provide feedback. AI Auto QA (Quality Assurance), which is expected to be introduced as early as the third quarter of this year, is a technology that arranges consultation content according to set criteria and provides consistent evaluation results. Once AI Auto QA is implemented, consultants will be able to receive feedback analyzed by AI based on their own consultation data.
In addition, LG Uplus plans to continuously upgrade the AI Consultation Advisor to increase the overall consultation time reduction rate to 30%. The company also aims to fully automate all non-consultation tasks based on an understanding of consultants, enabling them to focus on important work.
New features based on STT (Speech-to-Text) technology will also be introduced. STT technology converts conversations with customers into text in real time, helping consultants avoid missing any information. At the same time, AI uses this data to understand the context of the conversation and recommend appropriate consultation information in real time. In July or August, LG Uplus will add a feature to detect 'risk calls'?calls from customers requiring urgent handling or with high dissatisfaction. Here too, STT technology will be used to help team leaders quickly understand the consultation content.
The AI Consultation Advisor was built on 'Exaone,' the large language model (LLM) developed by LG AI Research. Sungkwon Jung, IT/Platform Build Group Leader at LG Uplus, said, "AICC is based on the 'Exaone 3.5 2.4B' model and utilizes 'Ixigen,' which is specialized for telecommunications. We are considering the simultaneous use of various large language models (LLMs) and will actively introduce new technologies in close collaboration with LG AI Research."
LG Uplus is also considering commercializing AICC as Software as a Service (SaaS) and providing it to enterprises in a business-to-business (B2B) format. The company currently provides AICC services, including callbots and chatbots, to more than 70 client companies. The AI Advisor is also being considered for service to LG Electronics client companies within this year.
LG Uplus is receiving questions and answers at the Artificial Intelligence (AI) Consultation Advisor briefing session. Namhee Seo, CV Manager; Sungkwon Jung, IT/Platform Build Group Leader; Jinhui Lee, New Service Development Lab Leader (from left). Provided by LG Uplus
© The Asia Business Daily(www.asiae.co.kr). All rights reserved.
