Real-Time Quality Review for 1.5 Million Monthly Consultations
Immediate Feedback on Use of Honorifics and Expressions of Empathy
LG Uplus announced on the 20th that it has developed an AI Auto QA, a large language model (LLM)-based coaching solution for customer service representatives, to upgrade the quality of phone consultations at its customer center. The photo shows an LG Uplus customer center representative receiving coaching. Photo by LG Uplus
On November 20, LG Uplus announced that it has developed and implemented an LLM-based counselor coaching solution called 'AI Auto QA' to enhance the quality of its customer center phone consultations.
AI Auto QA is a program that automatically reviews conversations and provides feedback immediately after a call ends. Previously, only a portion of the 1.5 million monthly consultations were evaluated by humans, leading to inconsistencies in fairness, and coaching feedback could be delayed by up to a week, limiting improvements in consultation quality. Differences in response quality among counselors had also been cited as a source of customer dissatisfaction.
To address these issues, LG Uplus introduced AI Auto QA. Counselors can now receive immediate feedback on detailed aspects such as the use of honorifics or expressions of empathy after each call. For example, if a counselor says, "There is a 5G plan available for you," the AI will correct it to, "There is a 5G plan available," and promptly notify the counselor.
According to the company, coaching staff can now delegate repetitive evaluation work to AI and focus on more advanced training, such as emotional coaching and sharing response know-how, thereby increasing coaching efficiency. New and less-experienced counselors can also quickly develop their skills through AI-based quality assessments and personalized coaching, which is expected to reduce service discrepancies among counselors.
AI Auto QA utilizes a proprietary LLM-based model that applies a 'Chain of Thought (CoT)' approach, evaluating all assessment items based on the flow and context of the conversation. While traditional rule-based methods assessed only single factors, the LLM can distinguish whether a customer's "Yes, I understand" signifies genuine understanding and acceptance or is merely a formal conclusion. The system also integrates a call noise detection model and voice frequency and prosody analysis technology, allowing for a multifaceted evaluation of consultation conversations.
Seo Namhee, Head of CV at LG Uplus, stated, "This technology has been patented and is currently being used for the evaluation and coaching of all counselors," adding, "We plan to offer it to external companies operating customer centers in the future to expand our sales."
© The Asia Business Daily(www.asiae.co.kr). All rights reserved.
