This deep dive expands on the Loop Asia conversation with Sutowo Wong, Managing Director, AI x Data at Temus.

If you've been handed the AI mandate — CEO, COO, VP Product — you know the tool isn't the hard part. The hard part is being accountable for an outcome inside an organisation with real infrastructure, in-flight commitments, and a board watching the clock. Sutowo Wong, Managing Director of AI x Data at Temus, built his apidays Singapore 2026 talk around that gap — two of his three "AI Triangle" conditions, sovereign and fast, speak directly to a Business Driver's position.

"Sovereign" doesn't mean what the RFP says it means

The obvious reading of data sovereignty is where the servers sit and what the contract says about access. Sutowo doesn't dismiss that — for much of the public-sector work Temus does, standard commercial arrangements with frontier model providers genuinely aren't adequate. But for a Business Driver, that's the easier half of the question.

The harder half is contextual sovereignty: your organisation already has infrastructure, in-flight investments, and people operating under real constraints. A solution that ignores that for a clean, greenfield rebuild isn't sovereign to your organisation — it's imposed on it. The test that matters: if your organisation can't carry the solution forward once the vendor leaves, it hasn't been transformed. It's been rented — exactly what a board directive to "deploy AI" misses, since it says move but not how. Sutowo's fix: build outward from the organisation's actual architecture and readiness, not inward from an idealised end-state.

The tempo mismatch is the real budget risk

Sutowo's third condition, speed, is where transformation programmes quietly lose the argument they thought they'd won. Model capability moves in months; enterprise deployment moves in years — a planning failure, not a technology one, and the one a Business Driver is best placed to catch, since it shows up first as a stale business case. The benchmark he cites — prototype in six hours, live deployment in six to twelve weeks — isn't a productivity boast, it's the tempo needed to close that gap. A three-to-five-year programme, measured against it, is a bet that the model landscape holds still long enough to catch up. Sutowo's view: it won't.

Why this has to be CEO-led, not CAIO-led

Sutowo's organisational claim lands directly on a Business Driver's desk: AI transformation can't be delegated to a Chief AI Officer — not because the role isn't useful, but because the work runs through business unit leads, who report to the CEO, not the CAIO. That means splitting each workflow into what AI runs autonomously, what stays human, and what AI assists, then reassigning training and reporting lines — authority a CAIO doesn't have. Sutowo's observation: when the CEO visibly owns the agenda, the organisation moves faster, because everyone knows they're being watched. That's the difference between a mandate that gets implemented and one that gets delegated until it stalls.

Governance as the thing that lets you move faster, not slower

The instinct in a board-mandated push is to treat governance as friction between mandate and deadline. Sutowo inverts that: what slows a programme is the absence of clarity about what's permitted. His analogy: you drive faster in a car you know has working brakes. Governance has to be designed in from the outset, not bolted on as a compliance review at the end.