3 days ago
- Business
- AU Financial Review
Agentic automation takes aim at Australia's productivity slump
'It augments things that you could not do with existing automation – especially things that cannot be defined in rules or direct approaches. It's more for dynamic goals, where agents can do a little bit of dynamic planning and decision-making along the way in order to perform the task.'
The recent UiPath Agentic Automation Summit in Sydney brought together business and tech leaders to showcase how agentic automation is already reshaping the way Australians work and engage with AI.
One summit attendee described agentic automation as a collision of technologies – a convergence of artificial intelligence and traditional automation that's already reshaping how businesses approach cost, quality and decision-making.
'Agentic automation is really the convergence of two trends,' says Dr Chris Marshall, vice president of data, analytics, AI, sustainability, and industry research at IDC.
'You've got generative AI, which is powerful but unpredictable and expensive, and traditional automation like RPA, which is reliable but limited. Together, they're becoming something more useful than either alone.'
Improved decision making
Marshall says this hybrid approach is helping businesses unlock a broader range of tasks that humans used to do. 'Suddenly you have a degree of reliability and intelligence that covers more work, more consistently, and often with better quality outcomes,' he says. 'It's not just productivity – it's also improved decisions and process quality.'
While the efficiencies are compelling, he says the longer term value comes from how agentic systems interact with companies' underlying processes. 'You start to learn where the blockages are,' says Marshall.
'You can simulate improvements, chop up tasks differently, or assign them to different agents or people. That's where innovation starts.'
He says the future of agentic automation isn't just advice from a bot – it's systems that act.
'The bot gets stuff done. The agent thinks so that people lead,' says Marshall. 'Combining advice, action and leadership – that's where agentic automation is heading, and that's what sets it apart from older forms of automation.'
Marshall says that Australia's approach to these tools may differ from other markets. 'Australians are more sceptical about technology – and that's not a bad thing,' he says.
'Especially in regulated sectors like financial services, that demand for privacy, data sovereignty and compliance is driving better questions and, ultimately, better outcomes.'
That's translating into real use cases. Businesses are starting to embed AI agents in the final stages of processes that were already largely automated, removing friction and extending the value of earlier automation programs.
While much of the conversation around agentic automation remains abstract, a recent deployment in Australia's energy sector shows what this technology looks like in action – and why businesses are moving fast to scale it.
Oil and gas take the lead
Resources giant Woodside Energy recently used agentic automation to overhaul its procurement communications, where staff had been manually sorting messages from more than a dozen channels. 'Employees were spending up to 20 hours a week just triaging communications,' says Peter Graves, area vice president ANZ at UiPath.
'With agentic tools, those hours are reclaimed, freeing people up to make decisions rather than sift through noise.'
The deployment used Communications Mining, a platform that applies natural language processing to unstructured inputs like emails and text messages and routes them intelligently across systems. 'Where traditional automation stops at structured systems, agentic automation bridges the gap into unstructured data,' says Graves.
By combining robotic process automation with AI and human input, the system enabled real-time decision-making while cutting delays and costs. 'It's about people, robots and AI agents working synergistically in the same workflow,' says Graves. 'That's the real shift.'
Woodside's implementation was powered by a bespoke dispatcher framework integrated with UiPath's CommPath language model. 'It's all run through a trust layer that makes the system secure and predictable,' says Graves.
Graves says the same architecture is already being replicated across other communication-heavy sectors. 'We're seeing this as the new standard for intelligent automation in Australia's energy sector and beyond,' he says.
'That's where the initial success is coming from,' says Graves. 'It could be a final step, or it could be a number of steps inside a process where you needed a human to jump in and make a decision. An agent can do that quite effectively.'
Recent IDC research shows strong Australian uptake, with more than half of organisations surveyed saying they are already using AI agents, and a further quarter plan to adopt them within the year.
In practical terms, this means fewer isolated bots and more systems capable of context-aware orchestration – taking in unstructured data, making decisions, and collaborating with humans across workflows.
The shift is particularly visible in sectors such as healthcare and financial services, where long processes and regulatory compliance create friction that traditional automation struggles to handle.
Unstructured data
'In healthcare, there are often human-in-the-loop steps that an agent can now manage,' says Graves. 'And in banking and financial services, there's lots of unstructured data. An agent can review, make a decision and move on – something it would have been impractical to code into a robot.'
This pressure is not just about productivity, but also about creating systems that are safe, governable and sustainable at scale.
While automation has long been associated with fears about job losses or burnout, early experiences with agentic tools suggest the opposite: by removing low-value cognitive load, they're helping people focus on more engaging work.
Removing the stress
'When I look at some of the things we have agents do right now, it's not typically the stuff people enjoy doing,' says Hao. 'It's the next level up from mindless work – stuff where you have to think about what to do next, but doing it isn't that interesting. It's almost always the same four steps, just ordered differently.'
Graves says that's having a measurable impact. 'We're seeing the ability for agentic automation to remove that stress, so someone can get their work done in the allotted time instead of staying back for two or three hours.'
Still, the shift is not without its barriers. Education, trust and clarity around what agentic AI can actually do remain issues for many organisations.
'It's still a very new area, and not a lot of workers have been exposed to it,' says Graves. 'There's a lot of proof of concepts, a lot of pilots – but not a lot that's gone into production yet.'
