BMMX

2.5802.38%

UOB to use A*STAR’s multilingual AI tool to guide SE-Asia relationship managers on client calls

LongbridgeAII'm LongbridgeAI, I can summarize articles.

SINGAPORE – UOB and A*STAR have tied up to explore how an artificial intelligence (AI) tool can help relationship managers better engage clients, as the bank drives greater deployment of AI across its operations.

The tool is designed to transcribe and translate customer calls in real time and prompt relationship managers with questions to ask and disclosures to make so they comply with regulatory guidelines.

Built on A*STAR’s large language model MERaLiON, the AI solution can understand 11 languages and dialects spoken in South-east Asia, where UOB’s retail customers are concentrated.

“(We) see potential for this solution to help our relationship managers better understand customers, navigate language and cultural nuances, and provide clearer, more relevant guidance,” said UOB’s group chief financial officer Leong Yung Chee in a statement on Nov 18 on the first day of the Singapore FinTech Festival.

In a demonstration to the media on Nov 13, UOB and A*STAR showed a mock customer call in which an early version of the tool could log sentences spoken in a mix of English, Mandarin and Malay.

When the relationship manager neglected to properly verify the client’s identity, the tool flagged it in red, prompted her to ask for an NRIC number and suggested a scripted line.

Panchai Aroonpairoj, a senior client advisor at UOB Private Bank, says the tool could come in handy when he speaks with older Thai clients who may sometimes use Teochew phrases during their calls.

An early demo version of the tool could transcribe sentences spoken in a mix of languages. A translation function will be built in later.

ST PHOTO: JOYCE ZK LIM

As he does not understand the dialect, he would otherwise have to ask them to repeat themselves in Thai, disrupting the flow of the conversation.

UOB believes that the solution could also provide insights for more targeted coaching of relationship managers, complementing existing training methods such as role-playing while a human observer gives feedback.

The bank hopes to roll out the AI tool sometime next year, though there is no fixed timeline as yet, said UOB’s head of enterprise AI Alvin Eng. For now, A*STAR and UOB will be working on developing the application and tailoring it to the bank’s needs.

Angela Chee, a director at A*STAR’s Institute of Advanced Intelligence and Computing, said this would mark MERaLiON’s first use case in the banking sector. Other banks such as DBS, OCBC and Standard Chartered have also been testing the model and evaluating how they would use it.

The model’s existing users include social service agency Lions Befrienders, which has a MERaLiON-powered voice agent to automate check-in calls on seniors, and tech service provider NCS, which uses the model to do voice transcription for its in-house chatbot.

A*STAR Director Angela Chee (left) and UOB's Head of Enterprise AI Alvin Eng at a demonstration of a new AI-powered tool for relationship managers.

ST PHOTO: JOYCE ZK LIM

“A large part of our work will be doing model adaptation to see how we can teach the model to speak or to think like a financial person, so (it) can transcribe more accurately,” said Chee.

A*STAR will have to familiarise MERaLiON with the terminology and acronyms that UOB uses, and also incorporate feedback from the bank as it tests out the solution.

UOB, like many banks, has been stepping up its deployment of AI to streamline workflows and drive value. It has more than 300 AI and data analytics use cases, including a solution that optimises cash replenishment routes and schedules for its ATMs.

The MERaLiON solution will not supplant its relationship managers. “That’s not our philosophy,” said Eng, adding that the tool is meant to augment staff.

For Aroonpairoj, the value of his team is clear. “You need relationships, which means you need human touch,” he said. “People will trust humans more than machines.”

Source: Straitstimes The copyright of this article belongs to the original author/organization.

The views expressed herein are solely those of the author and do not reflect the stance of the platform. The content is intended for investment reference purposes only and shall not be considered as investment advice. Please contact us if you have any questions or suggestions regarding the content services provided by the platform.