Australia enters the agentic AI era

Australian workers are using artificial intelligence more than ever, and the technology is changing how offices run. A new report from technology company HP found that 56 per cent of Australian knowledge workers now use AI tools at work every day or every week. That number jumped from 38 per cent just one year ago. Businesses across Australia are now asking a bigger question: what happens when AI doesn't just help people do their work, but actually does some of the work itself?

This new kind of AI is called 'agentic AI.' An AI agent can carry out a series of steps on its own, moving through different computer systems to get a job done. About 45 per cent of Australian knowledge workers say they already use AI agents, which puts Australia ahead of many other developed countries. That is a big change from the older way of using AI, where a person would ask a question and the AI would answer, but the person still had to do the rest.

Brad Pulford is HP's top manager for Australia and New Zealand. He says the shift is clear from watching how workers actually use the technology day to day. Many businesses know about AI, but they still need to help their teams understand why it is being introduced and what it means for their jobs. Getting workers on board with a clear plan is one of the biggest challenges companies face right now.

A separate report from consulting firm KPMG Australia backed up these findings. It found that 89 per cent of Australian organisations expected AI to stay one of their top investment priorities. It also found that 83 per cent of companies had their chief executive personally in charge of AI plans. This shows that AI has become a key business issue that leaders at the very top are now focused on.

One of the tricky parts of using AI agents is figuring out who is responsible when something goes wrong. When a worker uses AI to help write an email, they can still check it before sending. But when an AI agent carries out many steps across different systems, it becomes harder to know who is in charge. Businesses need to decide where AI's authority ends before problems arise, not after.

HP's research also found that AI use is not even across all workers. IT decision-makers use agentic AI much more than regular workers, who trailed behind by 16 to 19 percentage points. This gap suggests that companies need to do more to help everyday workers learn and use these tools. It is not enough for only the technology experts to be comfortable with AI.

Human oversight becomes more important as AI takes on harder tasks. HP's global research found that 78 per cent of skilled AI users felt confident they knew when to trust AI and when to use their own judgment. Only 52 per cent of beginners felt the same way. Pulford said that leaders should train workers to question AI results rather than just accept them without thinking.

Businesses also have to decide where AI does its computing work — on the device itself or on distant servers called 'the cloud.' On-device AI processes information right on the laptop or computer in front of the worker, while cloud AI sends it to powerful computers far away. Each approach has trade-offs involving speed, cost, and security. Pulford advised companies to think about the task and the data first, before choosing the technology.

The kind of computers workers use is also changing because of AI. HP's research found that 51 per cent of Australian knowledge workers said faster, more reliable devices would help them do their best work. Yet only about one-third said they had been given an upgraded computer to run AI, compared with about half of business leaders and IT managers. Companies will need to invest in better hardware if they want all employees to benefit.

Looking ahead, the big question for Australian businesses is not whether they will use AI, but what they will allow it to do. As AI agents move from helping one worker to carrying out tasks across whole teams and systems, companies must make clear rules about where machines act and where humans decide. Pulford summed it up simply: guardrails and good judgment are what turn AI adoption into something businesses can trust. Getting those rules right will shape the future of work in Australia.