KT will commercialize an "ultra-personalized artificial intelligence (AI) agent" tailored to users' preferences in the second half of this year.
Kim Jun-seok, head of KT AX Future Technology Institute's Agentic AI Lab (executive director), said on the 17th at the Gwanghwamun headquarters in Seoul that "instead of giving the same answer to every user question, we plan to apply an AI agent that provides ultra-personalized answers to KT's business-to-consumer (B2C) AI services in the second half of this year."
AI is evolving beyond the era of "chatbots" that simply answer users' questions into AI agents that think and act on their own.
KT plans to apply an ultra-personalized AI agent to the customer service app "My K" to analyze users' mobile usage patterns and conversations and design optimized plans. It will also equip the AI service "Genie TV," which performs content search and recommendations on the Genie TV set-top box, with ultra-personalized features to recommend more accurate, individualized content based on users' viewing habits and conversation histories. For example, if a user says "I want to watch a baseball game," the AI agent will recognize that the user is a KT Wiz fan and turn on a KT Wiz game.
KT said it will accelerate the expansion of industry-specialized vertical AI to target the business-to-business (B2B) market. Vertical AI under development by KT includes: ▲a network optimization agent that diagnoses network failures ▲a patent agent that preliminarily reviews patent filing potential ▲a legal agent that answers legal questions based on precedents ▲an easy-RFP agent that analyzes requests for proposal (RFP). KT aims to use these internally while also selling services to corporations and institutions.
Kim said, "The network optimization agent could also be used at power plants," adding, "We plan to bid for the 'sentencing support AI platform development project' being pursued by the Supreme Court with the patent agent."
KT is also pursuing research on next-generation AI agents. AI is evolving beyond merely telling users how to do something to actually completing work processes, such as clicking buttons or moving files. On this day, KT presented development cases for an Employee Agent that, when an office worker is away on vacation, has AI respond to colleagues' work inquiries and schedule meetings on the worker's behalf, and AI agents that set up golf outings by exchanging information with each other rather than through people.
Kim said, "Competition over foundation models is extremely fierce, and competition over agent technology among corporations is just as intense," adding, "KT is putting a lot of effort into improving the quality of AI agents."