DBS Upgrades AI Assistants to Support Banking Tasks for Over 10 Million Users

Singapore's largest bank is making a bet that feels both obvious and quietly ambitious: that the future of banking isn't a mobile app overhaul, but an AI agent that sits inside it, ready to act. DBS has spent the better part of a decade building AI capabilities — from fraud detection to personalised insights — and this week's upgrade to its virtual assistants marks a deliberate shift from passive information retrieval to active task completion.

DBS Joy, the corporate banking assistant, went fully agentic in Singapore earlier this week, giving 350,000 SME and corporate users the ability to complete real banking tasks within a single conversation. Customers can now ask whether a payment went through, review all transfers made to a specific recipient over a period, or check fees incurred — and the assistant retrieves, analyses and presents the data on the spot. DBS digibot, the consumer-facing equivalent, has also received generative AI upgrades since early this year, with new wealth-related conversational features due to launch next week.

The rollout spans three markets — Singapore, Hong Kong and Taiwan — and reaches more than 10 million users across both assistants. DBS plans to extend DBS Joy's agentic capabilities to another 100,000 corporate users in Hong Kong by September, with a phased rollout across other key markets. Beyond wealth, the bank intends to add agentic functions for common servicing tasks: checking card usage, reviewing rewards points, and blocking or replacing cards. The assistants are expected to handle more than one million chats each month.

There's a pattern here worth watching. Most banks treat AI as a bolt-on chatbot behind a 'talk to us' button. DBS is embedding agentic capabilities directly into authenticated workflows — the assistant only acts after login, and only on instructions from verified customers. That distinction matters because it opens the door to higher-stakes tasks: payments, account management, document handling. The walled-garden approach means DBS can deploy AI without the hallucination risk a general-purpose chatbot faces, because the assistant operates within a known data model with deterministic transaction outcomes. Customers can still escalate to a human service officer for complex requests.

Why it matters for Singapore: DBS is not just deploying a feature — it's defining an operating model for agentic AI in regulated financial services that other Southeast Asian banks will likely follow. When a bank processing millions of daily transactions commits to handling a million monthly conversations through autonomous AI agents, the implications ripple across the entire region's fintech landscape. For Singapore's Smart Nation ambitions, it signals that the country's anchor financial institution sees agentic AI as production-ready, not experimental — and that has a way of pulling the entire ecosystem forward, from MAS regulatory frameworks to the startups building adjacent infrastructure.