KT is using AI to standardise the design of app and web services and to automate drafting and reviewing customer guidance messages. Through AI transformation (AX), it is reducing repetitive development work and providing a consistent customer experience.

KT said on Nov. 12 it has built 'KDS (KT UX Design System) 2.0', a user experience (UX) design system with enhanced AI functions, and introduced an AI agent that supports drafting and reviewing customer guidance messages.

KDS is a common set of standards for screen layouts, design elements and guidance text applied to KT's digital services. KDS 1.0, released last year, focused on establishing design principles. KDS 2.0 is characterised by being advanced so that AI can directly use those standards.

KT built 'KDS Harness', which enables AI to compose screens and check outputs based on design standards. It reduced the burden of repetitive design work and revisions and increased consistency in screen layouts and usage methods by service. KDS has been applied to My KT, revamped in September, as well as KT.com, the Direct Shop and the AI model platform 'TokenFactory'.

TokenFactory in particular used KDS and ran planning, design and development in parallel, automating repetitive coding and documentation work. In processing a total of 424 functional requirements, 25 components and about 1.1 million lines of code, it completed in 2.4 months work that would have taken 8 months under the previous approach.

KT also introduced AI to customer guidance message drafting. KT sends about 170 million text messages each month to provide information on bills, service use and various benefits. The 'message generation agent', formally introduced internally on Oct. 30, drafts and reviews customer guidance text in line with KT's writing standards and relevant legal standards. It also provides image recommendations suitable for RCS texts that include images and buttons, and an automatic multilingual translation function.

The message drafting and review time fell to about 5 minutes from 30 to 60 minutes. Image production time was reduced to 15 minutes from 4 hours, and translation work that took 2 days or more was cut to 30 seconds.

KT also analysed customer responses by message composition. In pre-tests for sending messages converted to RCS, customer response indicators improved by about 1.3 to 3.4 times compared with the previous level. On Nov. 20 it plans to also unveil a 'customer language agent' that checks wording used across apps, the web and text messages and changes difficult terms into simpler expressions.

KT plans to expand the scope of KDS application to external partners, various digital channels and AI devices, and to further advance its AI-based design review system.

Won Man-ho (원만호), an executive director and head of KT's AX Platform Division, said, "With KDS 2.0, we are increasing design consistency and also shifting customer communication work to AX to improve work productivity and customer experience together." He added, "We will continue to broaden the scope of AX so that, no matter where customers encounter us, we can create a 'one KT experience' that is easy, consistent and reliable."