AI BPO technology company Ubase Group said on Tuesday it is stepping up efforts to commercialise its "AI Agent" service that directly performs customer counselling tasks.

The company said the Ubase AI Agent offers a range of functions centred on a conversational AI agent that analyses the intent and context in customers' utterances to support counselling, and a task-execution AI agent that directly processes requests such as automatic transfers, instant payments, delivery tracking and card loss reports. It uses retrieval-augmented generation (RAG) technology to search for and reference related knowledge such as up-to-date product and event information and FAQs to generate answers. It also provides personalised responses and product recommendations.

The Ubase AI Agent uses an open-source small language model (sLLM) to provide a conversational counselling engine that reduces the burden of application programming interface (API) costs. The company said it combines generative AI with domain-specific training in speech-to-text (STT) technology to identify a wide range of customer intent and context, achieving high speech recognition accuracy. It also applies ultra-fast processing with real-time factor (RTF) performance of less than 0.1 and deep learning-based high-quality text-to-speech (TTS) technology to provide voice counselling that converses naturally with customers, the company said.

After the counselling ends, AI automatically performs follow-up tasks such as summarising and classifying the counselling content.

Ubase CEO Mok Jin-won (목진원) said, "The Ubase AI Agent is a core technology that presents a new operating model for contact centres." He said, "We will implement a hybrid AI contact center (AICC) where AI and counsellors work together, and based on the AICC innovation capabilities and operational know-how that Ubase has accumulated, we will continue to strengthen our differentiated competitiveness."