Learn how to become AI-native by redesigning workflows with AI agents, building responsible AI foundations, and turning investments into productivity gains.
Esther Monks
Industry Advisor
Workday APAC
Agentic AI is fundamentally changing how work gets done. But realising that value requires organisations to think differently about work itself.
This was one of the messages that came through repeatedly in the latest episodes of the The Productivity Levers Vodcast series, produced by InnovationAus.
Here are some key insights from the vodcast—from rethinking how work flows and where governance sits, to matching human oversight with machine-scale autonomy.
Report
Why Agentic AI Changes the Productivity Equation
Most organisations are still thinking about AI in terms of assistance: summarising emails, generating text or helping employees complete tasks faster.
But there is a ceiling to that kind of productivity gain.
As Workday APAC CTO, Shan Moorthy, explains: “It’s only now that we’ve stepped into this agentic AI world where you can actually start delegating tasks…that’s where the real potential is.”
Take recruitment in healthcare. Today, a night-shift nurse applying for a role at 3am may have to wait hours for HR to log on and respond, creating delays, missed messages and frustration on both sides.
With an autonomous recruiting agent, the applicant gets an immediate reply. The AI agent can review their qualifications, confirm suitability and schedule an interview with a human recruiter.
When HR logs on in the morning, qualified candidates and interview schedules are ready and waiting—with the candidate details logically presented against the list of questions the interviewer chose to ask.
Shan also expects to see dramatic changes in any process that used to rely on employees applying policy, procedure and compliance information.
“When you onboard someone into an organisation, the first thing you do is hit them with a 6,000-page manual and say, ‘Here’s all the rules you have to adhere to, memorise them and keep them in the back of your mind while you do your daily tasks’,” he says, explaining that this is an impossible expectation.
In contrast, AI agents embed that knowledge directly into the workflow itself. Instead of expecting employees to carry the cognitive burden of compliance obligations, agents can shape human tasks in ways that align to policy and governance requirements.
Rita Arrigo experienced this first-hand in a dramatic change to onboarding when she joined Alchemy Impact as AI Strategy and Growth Lead.
“I didn’t have to watch a whole lot of videos and try and figure out how to do my expenses. [When I needed to do a task] I had an agent that answered my questions in the language of the organisation," she says.
As Rita discovered, work feels different—easier, faster, less frustrating—when layers of bureaucracy disappear behind the scenes, and an AI agent remembers and executes the compliance details for you.
In other words, bureaucracy is often created to mitigate risk, and AI is a tool that can help us reshape that.
The Hidden Risks of Autonomy
As a different type of productivity lever, agentic AI requires a different type of governance.
The advantage of agentic AI comes from its scale and autonomy. But those same characteristics also create a very different risk profile.
Where human mistakes tend to be relatively contained, slower moving and easier to correct, agentic AI changes that dynamic completely.
As Shan explains: “If I make salary changes and get a calculation wrong, it affects the 20 people that report to me. In that case, I can message them, ‘Hey, I didn’t get that right,’ and risk is mitigated. But if I give that same problem to an AI agent, it could affect thousands of people in five minutes. So the same risk has a huge blast radius.”
This is why organisations cannot simply layer AI agents onto existing governance models. The scale, speed and autonomy of agentic systems require a fundamentally different approach to oversight, testing and accountability.
This is particularly vital for any workflow that involves people or money. Because when it comes to HR and finance, almost right is wrong.
Responsible AI as the Bridge to Scale
Responsible AI is the operational and design foundation that allows organisations to trust AI systems enough to embed them into real workflows.
As Rita Arrigo explains: “Your AI isn’t just interesting and powerful, [it should be] governed, tested, monitored, accountable.”
Workday Public Sector Director, Paul Leahy, agrees. He argues that organisations cannot separate AI performance from the systems and data environments feeding it.
“You have to apply AI against a back office that has great data…where you know the context and structure and security and governance,” he says.
Legacy systems create a significant trust problem, because organisations often lack confidence in the quality, consistency and governance of the underlying data itself.
This is amplified when generalist AI tools are 'bolted on' over the top.
As Paul explains: “The challenge with some of these legacy systems is you're not sure about what data you're actually putting into the AI. Now is not the time to take risks. Now is the time to take actions that will underpin good AI. Legacy is holding organisations back from being able to do that properly.”
At the same time, organisations are confronting a different problem: having too much data.
AIIA CEO, Izzy Whitelock, warns that the rush to collect and store information can create its own governance risks.
“Organisations are gathering data at breakneck speed. Storage is cheap now, so we're gathering data left, right and centre. The question then becomes: what are we doing with that data?” she says.
Her point is that more data does not automatically create better outcomes. In fact, the sheer volume of information can obscure what matters most, reinforce hidden biases or create false confidence in flawed conclusions.
As Izzy notes: “Sometimes the volume of data blinds us to what the possible outcomes can be.”
Five Ways to Become AI Native
Becoming AI-native requires organisations to rethink how work flows, where accountability sits and how humans and AI agents operate together.
1. Start with outcomes, not technology
Define the business or customer outcome first, then work backwards to the process or technological capability required to support it (ironically, this may or may not include AI!).
2. Build an AI-native mindset
Encourage a new approach to problem solving by asking whether a workflow could be automated or redesigned around AI capabilities, before defaulting to traditional processes.
A practical way to frame this is to think of AI agents as a new team member, then ask what you would delegate to them.
For example, the examples Rita shared in the vodcast about summarising consultation documents and answering procurement queries are great use cases for AI, while judgment, relationships and ethical reasoning might be best led by people.
This requires both leaders and employees to understand how AI changes workflows, decision-making and productivity at a practical level—and that means getting people to engage directly with the technology itself.
At Alchemy, leadership training includes building their own AI agents so executives understand both the possibilities and governance implications of agentic AI.
3. Replace big-bang transformation with continuous iteration
Traditional multi-year transformation programs are too slow and too rigid for the pace of AI change.
Organisations need to shift to continuous experimentation, iteration and incremental improvement.
That means enabling teams to test smaller use cases in controlled environments, learn quickly and build trust progressively through visible wins.
Already, we can see the rise of smaller cross-functional 'fusion' teams focused on solving targeted business problems and demonstrating value quickly before expanding further.
4. Rebuild governance for an agentic world
Most of today's governance frameworks were built to mitigate human risk.
Agentic AI needs models built for autonomous systems—not frameworks that simply force AI to behave like a faster human worker, slowed down with human reviews.
That means rethinking oversight, escalation, testing and accountability for a world where AI agents participate directly in operational workflows.
It also shifts ownership from IT to the business.
IT can establish the foundational guardrails, but the business will own what "good" or "bad" behaviour looks like—much as it does with its human workforce today.
5. Match human oversight to machine autonomy
Not every agent needs the same level of supervision.
Low-risk, repeatable workflows may require minimal intervention, while high-impact decisions affecting customers, employees, finances or compliance obligations will come with tighter human review and escalation pathways.
The key is ensuring human agency appears in the right workflow, at the right time, while allowing AI agents enough autonomy to deliver meaningful productivity gains, in-line with organisational risk appetite frameworks.
Ultimately, becoming AI-native is less a technology project and more a leadership commitment and employee culture, enabled by capability.
It's about redesigning work intentionally, governing autonomy responsibly, and building the human and data foundations that allow AI agents to deliver lasting, meaningful productivity gains.
Watch all four episodes of the The Productivity Levers Vodcast now, produced by InnovationAus.
Report
A gap between effort and impact is the real ROI problem. Discover the solution.