Source: DBS Newsroom — 2026-07-28
Summary
DBS announced its generative-AI virtual assistants — "DBS Joy" for corporate and SME customers, "DBS digibot" for retail customers — have crossed 10 million users across Singapore, Hong Kong, and Taiwan, and that DBS Joy in Singapore has gone fully agentic: instead of only answering questions, it can now execute banking tasks like checking payment status, reviewing transfers, or checking fees within a single conversation. DBS reports active users up 61% and customer satisfaction up 17% in the first half of 2026, with the assistants together projected to handle over one million conversations per month. Agentic capability is next planned for DBS digibot in Hong Kong.
Key Takeaways
- The real shift isn't scale, it's capability: going from "answer a question about your account" to "execute the action the customer was asking about" is the difference between a chatbot and an agent, and it's happening inside a systemically important bank, not a startup.
- 61% growth in active users and 17% improvement in satisfaction in one half-year is a meaningfully large jump for a mature banking product, suggesting the agentic upgrade — not just marketing — is driving real usage.
- Rolling this out first to corporate/SME customers (higher transaction complexity, higher stakes) before retail customers suggests DBS is treating agentic banking as something to prove out on harder cases before mass retail deployment.
- The Hong Kong expansion for DBS digibot, still pending, is the next real signal to watch — it will show whether the agentic capability generalizes across regulatory environments and customer types.
Reel Script
Hook (~17s, 38 words): DBS just turned its bank chatbot from something that answers your questions into something that does your banking for you — and it's already live for over 10 million customers across three markets.
Core Concept (~65s, 150 words): The distinction that matters here is "conversational" versus "agentic." A conversational chatbot can tell you your account balance or explain a fee — it retrieves and reports information, but a human still has to go click the buttons to actually do anything. An agentic assistant collapses that: you ask it to check a payment's status or review a transfer, and it goes and performs that action itself, inside the same conversation, instead of just describing where you'd go to do it. That's the upgrade DBS Joy just got in Singapore for corporate and SME customers. It's a meaningful jump in trust and integration — the assistant now needs write-level access into banking systems, not just read-level access to answer questions, which is exactly why banks roll this out carefully and to business customers first.
Hands-On (~50s, 115 words): The numbers worth putting on screen: over 10 million total users across DBS Joy and DBS digibot, active usage up 61% and customer satisfaction up 17% in the first half of 2026, and a projected run rate of more than one million customer conversations per month. Draw it as a simple before/after: "chatbot answers questions" on the left, "agent checks payment status, reviews transfers, checks fees — inside the chat" on the right, with the usage and satisfaction numbers stacked underneath as the measured effect of making that jump.
Takeaway (~22s, 48 words): This is a real production proof point that agentic banking assistants can scale past pilot into tens of millions of users without the wheels coming off. If you're in fintech watching for when "agent" stops being a buzzword and starts being a deployed feature — this is that moment.