AI won’t deliver productivity without legacy systems reform

With persistent budget pressures and near-zero productivity growth, the Australian Government must adapt fast to do more with less. But this isn’t just a financial challenge: public sector productivity underpins national prosperity and directly impacts Australian living standards. When it stalls, so too do real incomes, service quality and the government’s ability to meet rising demand.

Unsurprisingly, AI has become the centre of that ambition. Across federal and state agencies, AI pilots are multiplying, bringing the promise of efficiency, faster delivery and reduced administrative burden.

But there is growing gap between the promise and the reality.

This is because, in practice, AI is being layered onto legacy systems without meaningful transformation of underlying processes, which 9 times out of ten doesn’t remove work, but simply redistributes it or even increases it. Instead of streamlined operations, agencies frequently come up against additional review steps, manual workarounds and duplicated efforts. This effectively creates new layers of digital red tape as staff attempt to resolve AI-generated outputs created on fragmented systems and inconsistent workflows — exactly the opposite of its desired effect.

In other words, government agencies are trying to use AI to compensate for complexity that existed long before it’s arrival, and that it cannot resolve on its own.

AI cannot be treated as a bolt-on to existing systems. The systems must be updated first for AI to be integrated properly, and unfortunately, government will continue to face challenges if it fails to do so. No matter how advanced the tool is, it will be constrained and inefficient if forced to operate with the low quality and siloed data present in legacy environments and ultimately expected productivity gains will remain out of reach.

Fixing the foundations

It’s clear that Australia must get serious about lifting public sector productivity, to protect both financial stability and living standards. AI can be part of the solution, but only if government face legacy complexity head-on. This means simplifying processes, unifying decision-making across channels, and ensuring data and workflows are orchestrated in real time, not stitched together through disconnected systems.

Done right, AI doesn’t just automate processes, it can remove the layers of red tape within them by simplifying workflows, reducing approvals and embedding compliance into the work itself.

However, adding another layer of technology, without integrating, only adds to an already complex environment, whereas rethinking how work flows across an organisation is when real change in productivity will be realised.

From pilots to impactful organisation-wide transformation

Once legacy constraints are addressed, AI can scale from isolated pilots to enterprise capability. However, in a risk-sensitive government setting, it can be easy to keep pilots isolated from core systems and heavily governed, which ultimately cannot deliver the impact needed.

To move beyond this, agencies must build on the foundations of their legacy transformation efforts, which will enable them to design for scale at the outset. This means connecting policy, operations and delivery, so work can flow end-to-end rather than being broken across disconnected systems.

Governance is also important. Risk, compliance and transparency are of course non-negotiable in the public sector, but that doesn’t mean it has to stunt innovation. When oversight is designed into how work happens, rather than layered on afterwards, agencies can move faster without sacrificing trust.

Finally, it is necessary to shift priorities from use cases to outcomes. Rather than focusing on isolated tasks, the emphasis should be on workflows across the organisation as a whole. This means reimagining the entire service journey to remove handoffs, duplications and delay across the value chain for true productivity at a system level, not just within individual activities.

A new model for public sector productivity

This is what ‘doing more with less’ looks like in practice: not adding more tools but removing friction and transforming workflows across the entire organisation.

The opportunity is huge and if done well, AI can help the government meet rising expectations, improve responsiveness and ease fiscal pressure. But only by tackling the underlying complexity of legacy systems can it improve outcomes.

AI must be part of a broader transformation of how work is designed to be successful, and in the current climate, failure to do so is not a risk government can afford to take.

Image credit: iStock.com/tadamichi