Share with your CHRO
Across Southeast Asia, the AI scaling gap in HR transformation is now quantified and it’s stark. Forty-three percent of SEA organisations have moved past AI pilots, but only 25% can scale to enterprise-wide deployment, and just 10% of HR and business leaders can demonstrate measurable ROI on their AI investments. A People Matters webinar featuring Hans R. Montenegro of Meralco and Mario Jacober of Zurich Insurance surfaced a four-part framework, centred on process re-engineering, data unification, single-workflow proof-of-concept, and cross-functional ownership, as the practical bridge across that gap.
What this means for your business
The 43-to-25 percent drop isn’t a technology problem, it’s a sequencing problem. Most HR functions that stall between pilot and scale have layered AI onto processes they never fixed, fragmented data they never unified, and ownership structures they never clarified. If your organisation sits in that gap, the diagnosis is almost certainly one of those three, and the number most worth interrogating is the 10% ROI figure, because it signals that the business case was assumed rather than constructed.
The sequencing argument the webinar makes is correct, and it cuts against the way most enterprise AI investments actually get approved. Budget cycles reward ambition, so leaders pitch AI transformation before they’ve mapped the underlying process, which means the technology arrives before the data infrastructure can support it. The agents-on-bad-data problem (deploying AI tools, essentially autonomous software that acts on information, onto fragmented or unreliable data sources) doesn’t just produce wrong outputs, it actively erodes trust in the entire transformation programme. Rebuilding that trust costs more time than fixing the foundation first would have.
The single-workflow-first model is the sharpest practical insight here, and it reframes a decision most CHROs already own, specifically which workflow they chose for their first AI deployment and whether it was actually end-to-end redesigned or just AI-augmented at one step. That distinction determines whether the organisation builds a repeatable scaling template or an expensive one-off. I’d revise this view if organisations with fragmented multi-workflow rollouts consistently achieved scale, but the 25% figure suggests they don’t.
Concept deep-dive: Process re-engineering before AI
Process re-engineering means redesigning a workflow from scratch around its desired outcome, before selecting any technology to support it. The analogy is renovating a building’s plumbing before installing smart fixtures; skipping the first step means the fixtures surface the problem rather than solve it. In AI deployments, this matters because the model or agent will faithfully execute whatever process it’s given, so a broken process becomes a broken process running faster and at greater scale.
Based on reporting from Webinar: Beyond AI in HR, Foundations for Org Transformation, SHRPA SEA insights, originally published 2026-10-21 00:45:00.
