Stop automating broken processes or face work slop

Keynote and panel reveal the dysfunction trap

HR Tech Asia 2026 (May 4–7 in Singapore) opened with a deliberate warning from Jason Averbook, co-founder of Now to Next: layering AI on top of broken processes creates "work slop"—bad data, bad processes, bad creativity. Averbook argued that HR has spent decades moving the same tasks from mainframe to cloud to AI without redesigning the work itself.

He framed this as structural: roughly 70% of HR work is execution (hands work), the category most vulnerable to automation. Rather than a threat, he positioned this as an opportunity to redirect time to strategy and human connection. But only if the underlying processes are sound.

A CHRO panel featuring leaders from Bank BRI, Press Metal, KPJ Healthcare, Central Retail Corporation, and Singapore's Public Service Division reinforced the point. Suryo Sasono from Bank BRI shared a concrete case: the bank deliberately chose not to fully automate credit decisions despite having sufficient data. "When you start outsourcing decision-making purely to AI, it becomes no one's responsibility," he said. "There needs to be one person accountable."

Press Metal's Andrew Chan described a multi-year shift from tenure-based to data-driven promotion decisions via a transparent talent dashboard—work that earned the company a UN Global Compact Award. Low Peck Kem of Singapore's Public Service emphasized that talent mobility requires deliberate design, not just consent: she noted that HR directors in the Singapore public service rotate sectors every five years specifically to normalize movement across the organization.

Accountability gets eroded or preserved by design

The conversation exposed a second-order problem: AI adoption metrics have become the goal, when embodiment should be. Averbook pushed back on simple adoption counts, citing McDonald's use of WhatsApp as a front-end to Workday as a signal of real cultural integration, not just tool deployment.

The accountability angle is the sharpest one. Once a decision is delegated to an algorithm without a named owner, responsibility evaporates. Bank BRI's example shows this isn't academic: the organization consciously preserved human judgment over credit decisions specifically to keep accountability intact. This matters because trust in HR systems erodes the moment employees believe no one is responsible for a bad outcome.

The panel also surfaced that leadership capability is hard to grow. When asked which is hardest to cultivate—curiosity, empathy, or systems thinking—panelists split. Panchalee Weeratammawat (Central Retail Corporation) flagged empathy as neglected; Sasono chose curiosity, warning that even the largest and most profitable institutions (he cited Bank BRI) face complacency risk. Dr. Tanvi Gautam closed with a pithy verdict: "A right person with the wrong technology could work. But the wrong person, even with the right technology, is not going to work."

Start with process audits, not tool pilots

The pattern across all three speakers and panelists is identical: before adopting an AI tool, identify which of your workflows are actually broken. Averbook's framework is useful here: separate execution work from strategy work from people work. Automation speeds up execution; it doesn't fix bad execution.

Second, assign accountability. Don't let AI become the reason no one owns a decision. If an HR system will influence hiring, promotion, or compensation, one person must be accountable for the outcome, even if the tool surfaces the recommendation.

Third, design for embodiment, not adoption. Ask whether your employees naturally interact with the system as part of how work gets done, or whether they resent it as a compliance step. The difference is the difference between a tool that works and one that doesn't.