Cover Bidyut Dumra, group head of Innovation & Future of Work at DBS and Oliver Jay, managing director, International at OpenAI, engaged in a conversation with Tatler’s Chong Seow Wei on the transformative potential of AI, measuring ROI and the importance of leadership in an increasingly AI-powered world (Photo: Adrian Lee)
Bidyut Dumra, group head of Innovation & Future of Work at DBS and Oliver Jay, managing director, International at OpenAI, engaged in a conversation with Tatler’s Chong Seow Wei on the transformative potential of AI, measuring ROI and the importance of leadership in an increasingly AI-powered world (Photo: Adrian Lee)
1 / 1
OpenAI’s Oliver Jay and DBS’ Bidyut Dumra shared sharp lessons on rolling out AI, measuring ROI, safety and winning over sceptical staff at a Tatler House Dialogue held in partnership with DBS Corporate & Institutional Banking
What does it actually take to make artificial intelligence (AI) work for a business that has spent decades doing things one way? That was the question hanging over Tatler House on the evening of September 22, as next-generation business owners and family business leaders gathered for a conversation that skipped the usual hype.
The Tatler House Dialogue, presented in partnership with DBS Corporate and Institutional Banking, brought together Oliver Jay, managing director, International at OpenAI, and Bidyut Dumra, group head of Innovation & Future of Work at DBS, moderated by Tatler’s Chong Seow Wei. Welcome remarks came from Chen Ze Ling, group head of Corporate and SME Banking at DBS, who welcomed a new generation of business owners bringing fresh instincts to the businesses their families built.
Above Chen Ze Ling, group head of Corporate and SME Banking at DBS (Photo: Adrian Lee)
Chen Ze Ling, group head of Corporate and SME Banking at DBS (Photo: Adrian Lee)
1 / 1
Redesign, don’t bolt on
For Bidyut, the tougher problem isn’t the technology, but the instinct to leave a working process alone. He told the room that DBS’ transformation moved from early experiments with AI in 2014 to a full bet on “industrialising AI” from 2018. The bank grouped its data professionals into one business unit to retain talent, trained managing directors to ask the right questions of new use cases, and tied 20 per cent of performance scorecards to transformation, not just output. His advice: “Put a heavy hand now, rather than later.”
Above Bidyut Dumra, group head of Innovation & Future of Work at DBS and Oliver Jay, managing director, International at OpenAI (Photo: Adrian Lee)
Bidyut Dumra, group head of Innovation & Future of Work at DBS and Oliver Jay, managing director, International at OpenAI (Photo: Adrian Lee)
1 / 1
You are what you measure
The bank also built an “AI economic value” framework, running every new use case as a test group against a control group to track the real gain in cost, revenue or productivity, a discipline disclosed in DBS’ annual report since 2022. One example he gave was how the bank used an AI model that spotted customers in India using high-interest credit cards who’d be better off on an instalment plan with lower interest rates, and nudged them toward it—a move that doubled the business.
But metrics cut both ways, he warned, adding that you have to be clear on exactly what behaviour you’re rewarding and make sure it’s being measured.
Read more: Tatler House Dialogue: Building family businesses that last with DBS Treasures Private Client
Start with the outcome, not the tool
Jay’s advice for leaders without a dedicated AI team echoed a theme he returned to all evening: start with the business outcome, not the technology.
Companies that tally up scattered productivity gains—a 5 per cent lift here, 20 per cent there—struggle to show real return on investment (ROI). Businesses that start with a concrete target, such as cutting a production cycle by two months, back into the right AI investment far more easily.
He pointed to Singapore’s hawker centres as proof of how accessible the technology has become, describing stallholders tracking ingredient costs and analysing customer preferences with more sophistication than plenty of large enterprises manage.
Guests asked Jay and Bidyut questions after the panel (Photo: Adrian Lee)
1 / 3
Safety is the real bottleneck now
If capability was the story four years ago, Jay said, safety is what now separates companies that scale AI from those that get burned.
OpenAI runs extensive “red teaming” to stress-test its models and build guardrails before release—protection that vanishes once a company downloads an open-source model stripped of its safeguards to chase performance.
His advice: treat the choice of AI partner as seriously as any other vendor decision, and ask difficult questions about the safety testing behind whatever platform you adopt before you do.
Tyler Ryu, engineer at OpenAI, doing a demo of ChatGPT Work (Photo: Adrian Lee)
1 / 2
The evening’s most vivid illustration came after the panel, when OpenAI engineer Tyler Ryu built a live business pitch using ChatGPT Work, sourcing costs, drafting a spreadsheet and building an investor deck for a fictional luggage-tag start-up.
Closing the evening, Soh Wee Li, head of Corporate Banking at DBS Singapore, urged the room to treat AI adoption as urgent rather than optional, pointing guests to DBS’ Spark GenAI Programme for tailored guidance.
Above Soh Wee Li, head of Corporate Banking at DBS Singapore (Photo: Adrian Lee)
Soh Wee Li, head of Corporate Banking at DBS Singapore (Photo: Adrian Lee)
1 / 1
Guests then moved to a networking session with food by Tim’s Fine Catering, a sake tasting by Dassai and a live leather-stamping station by Bynd Artisan, where guests personalised their own take-home tags.
Click through the gallery for more event photos.
Guests of DBS Corporate & Institutional Banking (Photo: Adrian Lee)
1 / 8
