KT Reduces Average Customer Consultation Time by 5 Seconds with Proprietary AI "Mi:um"

Applied "Mi:um" to KT and Group Customer Centers
Consultation Content Suggestions and Post-Consultation Automation
Consultation Result Organization Work Reduced from 40% to 3%

Real-time consultation chat and real-time chat monitoring screen applied at KT Customer Center. Photo by KT

On July 30, KT announced that it has officially implemented its self-developed generative AI model "Mi:um" in its customer service center operations, resulting in an average reduction of five seconds per consultation per agent and improving customer service convenience.

Since 2023, KT has been applying its AICC (AI Contact Center) solution, based on its proprietary language model "Mi:um," to customer service centers across major group affiliates, including KT itself, K Bank, and KT M Mobile. Mi:um is characterized by its ability to effectively support operations with only a small amount of consultation data, thanks to its extensive pre-trained vocabulary data.

The "AI Consultation Assist" system analyzes customer inquiries in real time and recommends appropriate responses. For example, when a customer asks about a "KT mobile plan," the AI displays relevant keywords on the screen, and when the agent clicks on them, they are immediately connected to related documents. This enables quick and accurate responses, increasing satisfaction for both agents and customers.

After the consultation, the system automatically summarizes and classifies the conversation. Agents can review the progress of the conversation and the customer's emotional state.

The effects of the implementation are evident in the numbers. According to KT's research, the average consultation time per customer has been reduced by five seconds, and for all wired and wireless customer service staff, a total of 4,500 work hours have been saved per month. In particular, the time spent on post-consultation result organization has dropped sharply from 40% of total work to less than 3%. KT explained that the additional time gained is being used for customer care activities, leading to improvements in consultation quality and customer satisfaction.

These achievements were reflected in KT being selected as a "Korea Excellent Call Center" for 14 consecutive years in the call center category of the Korea Service Quality Index (KSQI), hosted by Korea Management Association Consulting (KMAC) in May this year.

KT is currently working to apply Mi:um to its "AI Voicebot" services, which are operated at customer service numbers 100 and 114. Starting in the second half of this year, KT plans to expand the application of Mi:um 2.0 across all customer consultations to further improve consultation accuracy and operational efficiency. In addition, KT aims to lead digital transformation projects for customer service centers using AI large language models (LLM) for a variety of clients, including those in finance, public sector, and healthcare.