DBS Bank’s AI Playbook: 430 Use Cases, 2,000 Models, Human-in-the-Loop
DBS Bank has turned AI into a core part of how it runs. The bank’s AI playbook now covers 430 use cases and more than 2,000 models built across the group. AI, short for artificial intelligence, is technology that lets computers learn and make decisions. A “use case” is simply a real task where AI is put to work. DBS is one of Asia’s largest banks, based in Singapore. Its story shows how a big, careful business can use AI at scale without losing the human touch.
The key word here is balance. DBS leans on machines for speed, but keeps people in charge of the big calls. This “human-in-the-loop” approach is the heart of its plan.
AI at a huge scale
The numbers are striking. DBS has delivered 430 AI use cases across the group. It has built more than 2,000 models. These cover retail banking, corporate banking, operations, and staff productivity. “Retail banking” means services for everyday customers. “Corporate banking” serves large companies.
In 2025, these data and AI efforts generated about S$1 billion in economic value. “Economic value” means real money gained or saved. DBS does not just guess this figure. It compares results from AI-powered services against control groups. A “control group” is a set of customers who did not get the AI feature. The gap between the two groups shows the true benefit.
As Saurabh Mittal, Head of Transformation and Data at DBS Bank India, put it, “We recognised that to harness AI meaningfully, it needed to be embedded into the organization’s fabric.”
Keeping humans in the loop
DBS does not let AI run alone for risky decisions. For complex credit, fraud, and risk choices, a human stays in the loop. This means a person checks and approves the AI’s work before action is taken. The bank calls its vision an “AI-enabled bank with a heart.” The idea is to mix machine speed with human judgement and care.
To get there, DBS rebuilt many of its workflows so people and AI work side by side. It also reskilled staff using a “Triple E” framework. The three Es stand for Education, Exposure, and Experience. In plain terms, staff learn about AI, see it in action, and then use it on real work.
The tools and the guardrails
DBS built its own set of AI tools. Here is what they do in simple words.
- ADA (Advancing DBS with AI) — the bank’s main data platform.
- GenAI Flock — a gateway that lets teams safely use generative AI.
- DBS-GPT — the bank’s own AI assistant for staff.
- CodeBuddy — a tool that helps with coding and analysis.
- ALAN — an AI protocol and knowledge store.
To keep AI safe, DBS uses the PURE framework. PURE stands for Purposeful, Unsurprising, Respectful, and Explainable. Every AI use case is checked against these four points. “Explainable” means the bank can show why the AI made a choice. This is how DBS practises “responsible AI,” which means using AI in a fair and safe way.
Key facts
| Detail | Figure |
|---|---|
| AI use cases | 430 across the DBS Group |
| Models built | More than 2,000 |
| Economic value (2025) | About S$1 billion |
| Total income (Q1 2026) | SGD 5.95 billion (about $4.6 billion) |
| Net profit (Q1 2026) | SGD 2.93 billion (about $2.27 billion) |
| Total assets (Q1 2026) | SGD 935.4 billion (about $725 billion) |
FAQ
What does “human-in-the-loop” mean?
It means a person stays involved in AI decisions. For risky tasks like credit or fraud checks, a human reviews and approves the AI’s work before any final action.
How does DBS measure AI’s value?
It compares AI-powered results against control groups who did not get the AI feature. The difference shows the real gain. In 2025, this came to about S$1 billion in value.
What is the PURE framework?
PURE stands for Purposeful, Unsurprising, Respectful, and Explainable. DBS checks every AI use case against these four points to make sure the AI is safe and fair.
Why it matters (especially for India and founders)
DBS gives a clear blueprint for AI done right. It proves that AI can add real money, not just hype. The S$1 billion figure is tied to measured results, which builds trust.
For Indian banks, startups, and founders, the lessons travel well. Build your own tools. Keep humans in charge of risky calls. Measure value against a control group. And put safety rules in place from day one. This kind of careful scaling is what turns AI from a demo into a business. To understand the wider chip and tech wave powering this shift, see how Micron posted record revenue. And for a look at AI’s next frontier, read how the first humanoid robot maker went public.
The takeaway
DBS Bank has shown how to scale AI with care. With 430 use cases, more than 2,000 models, and about S$1 billion in value, the results are real. Yet people still hold the wheel for the hardest decisions. That mix of machine power and human judgement is the playbook other banks and founders can learn from.
Sources
- Analytics India Magazine — DBS Bank’s AI Playbook: 430 Use Cases, 2,000 Models, Human-in-the-Loop
- Computer Weekly — DBS rewires operating models for AI reasoning era
Related coverage
- China Will Have a Mythos-Class AI Model by Year-End, Says Z.ai Founder
- Kerala Bets Big on AI, GenZ Startups and Geospatial Intelligence in Revised Budget 2026
Get the day’s top stories in your inbox
One concise email. No spam, unsubscribe anytime.
