The company serves more than 11,500 organisations worldwide, including over 65 per cent of Fortune 500 companies, with operations in more than 175 countries.

Beyond pilots and proof of concepts

bootstrap slideshow

AI should be viewed as a tool that enhances productivity, allowing employees to focus on higher-value tasks instead of repetitive processes, she added. In finance, that could mean agents handling routine reconciliation so staff can focus on anomalies rather than every line item. In recruitment, this could involve screening large volumes of CVs to surface the strongest candidates.

“For the most part, customers that work with Workday already have a strong foundation where they can understand their workforce and their finances. Now, through the AI agenda, it's how we build agents and AI on top of that foundation to make them better at what they do.”

With that foundation in place, O’Reilly said many Malaysian enterprises are looking to take the next step by building AI capabilities on top of existing human resources and finance systems rather than replacing them altogether.

“AI still sits in pilots and proof of concepts. There’s a big desire, particularly from senior leaders, to move to something far greater that has bigger ownership and bigger impact,” she said.

As organisations begin deploying AI beyond pilots and into business-critical functions, however, trust and governance become even more important.

Why trust comes first

O’Reilly stressed that trust must remain at the centre of AI deployment. While AI is probabilistic by nature, critical business functions such as payroll and finance require deterministic foundations supported by governance and safeguards.

Arguing that probabilistic AI needs to be anchored by a trusted, deterministic layer of data and controls, O’Reilly said, “A payroll can't be ‘maybe okay’. Reconciling at month-end can't be $10 off or $100 off.”

O’Reilly framed this as the difference between what Workday calls “lawful” and “lawless” agents. Lawful agents operate within established permissions and controls. This is opposed to lawless agents that can access information without boundaries.

“AI pessimists have the right to be pessimists because they’re probably thinking about the deployment of AI agents in a lawless environment. But when you’re building on foundations with guardrails, with trust, access and controls, then you should be able to have a load of trust in what you’re deploying.”

Across the industry, many enterprises invest at the front door - the user-facing tools and interfaces -- while leaving the underlying foundation unresolved.

Getting AI right requires investment in the core: trusted, well-governed data brought together in one place, so that agents operate on a reliable single source rather than fragmented systems.

Governing humans and agents together

O'Reilly said AI and humans would increasingly work side by side, requiring organisations to govern digital agents much as they manage employees today. Onboarding an agent, she suggested, should mirror onboarding a new hire, who is granted an employee ID and given access and permissions appropriate to their role.

She said Workday's vision of an “agent system of record” would allow organisations to oversee both human and digital workforces in a single view, measuring performance and ensuring accountability across both.

On concerns that AI could replace jobs, she said humans would remain responsible for decision-making and governance, with AI acting as an enabler rather than a substitute.

O’Reilly pointed to a shift from "human in the loop", where a person signs off on every action, to "human on the loop", where people govern and monitor the flow of work without intervening at every step.

“Humans are always responsible, which means we have to play a big role in how we architect, how we govern and how we monitor what happens.”

Culture and enablement as the missing piece

O’Reilly highlighted the importance of creating a culture that encourages AI adoption from the top, saying leaders should expect employees to use AI tools to improve productivity and analysis rather than treat them as optional extras.

She described expecting her teams to arrive at meetings having already used AI to gather insights, so discussions can focus on analysis rather than gathering preliminary insights.

While optimistic about Malaysia's prospects, O'Reilly said organisations need to invest more heavily in training and enablement to fully realise AI's potential. This speaks to the hurdles the government has flagged in its approach to AI -- from skills gaps to limited governance oversight -- where a trusted, deterministic foundation can bring reliable, well-governed AI within reach for the medium and growing enterprises most at risk of being left behind.

She noted that bridging this enablement gap is critical as Malaysia rapidly cements its position as an attractive regional hub for finance and technology, drawing an increasing number of international operations and outsourcing activities.

“There is a lot of expectation for Malaysia to step up and be able to deliver on this promise.

“When all of this demand around being a finance and tech hub happens, how do we manage the technology, the governance, the risks? There are standards that need to be met and there are going to be lots of eyeballs on that.”

Unlocking this potential, she said, requires businesses to move away from viewing technology as a threat to job security and toward exploring AI agents as tools to achieve the growth expected in today's market.

In Southeast Asia, Workday has established a presence across six markets, including Malaysia. Its Malaysia-headquartered clients include AirAsia, BPMB and Sime Darby.

The company has also expanded its AI capabilities through agent-based technologies designed to automate routine tasks and enhance decision-making across enterprises.

-- BERNAMA