AFEM

85.9750.16%

Banks prepare for AI agents to take over the customer front end

LongbridgeAII'm LongbridgeAI. Here is the key information I extracted
- Executives from Tencent, UOB, HSBC, and Visa stated at the Singapore FinTech Festival that AI agents are expected to take over the customer front end in banking, with 75% of customers wanting AI assistance in their banking relationships.
- BCG research indicates that while 23% of customers already use AI for banking, 82% still want to approve every financial decision, prompting financial institutions to develop phased controls, guardrails, and trusted agent protocols.…
Disclaimer

Four in five want AI and agentic assistance built into their main banking relationship.

Banking apps will remain but move into the background as customers’ AI agents take over how they discover and choose financial products, executives from Tencent, UOB, HSBC and Visa said at the Singapore FinTech Festival.

The shift could leave banks with fewer direct chances to earn customers’ attention and trust, and the panellists said institutions will need to rethink their channels, service design and how they handle customer consent.

“Customers are going to use their personal agents. There is no question about it,” said T.R. Ramachandran, head of products and solutions for Asia Pacific at Visa.

Ramachandran said the question isn’t whether customers use banks or agents, but where the journey begins and ends. He expects it will increasingly start with a customer’s agent discovering banking products, whilst the transaction still culminates in a bank’s app or digital channel.

Daniel Hong, vice president of Tencent Financial Technology, said banking apps will go the way of physical branches, still present but with a different role.

Apps will become the back end that handles the heavy lifting, whilst agents serve customers at the front end through natural language.

“Users don’t have to understand the menu, don’t have to figure out the buttons,” Hong said.

Customers already appear ready for the change. BCG research found 23% of its customers already use AI for their banking needs, and 75% want AI and agentic assistance built into their main banking relationship.

However, 82% said they would still want to approve every financial decision an agent makes, and the research found customers are less comfortable handing high-stakes, emotional decisions, such as mortgages and investment management, to AI.

Tencent’s answer is to let customers hand over control in stages. Its AI agent payment product, launched on WeChat this year, uses an isolated sub-account that only agents can spend from.

In the first phase, every agent payment needs the user’s approval and password. In the second, users only confirm their intent. In the third, agents act on their own within a set task and budget, such as planning a trip under US$2,000.

“The key to building trust in agents is not to ask for the trust, but to set up an environment so that the user can test, explore,” Hong said. Tencent is also working to make more mini programs agent-ready.

For Lee Zhu Kuang, managing director and head of UOB’s innovation group, the challenge is that both customers and banks will use agents, so institutions must decide how much authority each one gets.

“When we give the authority to an agent, it must reflect the customer intent, the impact of the decision as well as the risk that is being involved,” Lee said. After this, banks can create the guardrails around it.

Lee said banks can earn that delegation by giving customers positive, controlled experiences with agents, and by using agents to support people in complex, regulated or high-value decisions.

Banks can also stand out by redesigning the customer journey, not just the interface. UOB is exploring a conversational chatbot that helps customers find flights and hotels and pay with either their card or reward points.

Ultimately, Lee said, the relationship still matters.

“How do we make sure that the clients have someone or an institution to turn to when things go wrong?” Lee said.

Shayan Hazir, chief digital officer for Asia (excluding Hong Kong) and MENAT at HSBC, compared deciding which tasks to hand to agents with aircraft autopilot.

Machines handle certain tasks safely within guardrails, humans take over once thresholds are crossed, and everything is recorded. Hazir expects that balance to shift over time, just as autopilot evolved over a century.

Hazir argued that agents won’t reduce banks to a mechanical comparison. Instead, he said, this is the best opportunity yet for banks to differentiate, moving from hyper-personalisation to products built for a “segment of one,” where customers shape their own investment portfolios, facilities and rewards.

HSBC recently completed a business-to-business transaction using agents with two clients, a technology partner and a payments partner.

“It’s not human in the loop; it’s human judgment in design that needs to be the case,” Hazir said, adding that banks must work alongside regulators, such as the UK’s Financial Conduct Authority, which has issued principles on agentic commerce.

Ramachandran said trust in agentic payments rests on identity, authentication, consent and authorisation, plus a new element, signals, which capture whether an agent understood the customer’s intent. That matters for resolving disputes and stopping agents from going rogue.

Merchants also need to adapt, since online stores are built for humans, not machines. Ramachandran said many merchants are either letting all bots in or blocking them entirely.

Visa created its Trusted Agent Protocol so merchants can confirm an agent is who it says it is and has the customer’s consent.

“Like-minded players need to come together to create the right environment, the right sort of ecosystem, and the right guardrails,” Ramachandran said, adding that customer adoption is where agentic commerce can succeed enormously or fall flat on its face.

Source: Singapore Business Review 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.