AI is now part of almost every public sector technology conversation. But government buyers are sending a clear message to the market: they are not looking for hype. They are looking for practical value.
Across recent government briefings in New South Wales, Queensland and Western Australia, the same themes kept coming through. Government wants AI that improves services, supports public servants, strengthens decision-making and reduces manual work. But it also wants AI that is safe, governed, secure and connected to existing public sector systems.
That distinction matters.
The opportunity for vendors is real. But the strongest opportunities will not come from generic AI pitches. They will come from understanding how government works and solving specific problems within that environment.
Government wants outcomes, not AI theatre
The best AI conversations start with the problem, not the product. Government leaders are asking practical questions:
- What service problem does this solve?
- What data does it use?
- How is risk managed?
- How does it fit with existing platforms?
- Can staff use it safely?
- How will success be measured?
These are the right questions. AI adoption in government should not be judged by how many tools are deployed. It should be judged by whether services improve, staff are supported and public trust is protected.
Queensland made the point directly: “Without that infrastructure in place, without sound governance of data, you’ll get nothing.”
The foundations matter
One of the clearest messages from government is that AI depends on the basics.
Good AI needs good data. It needs secure infrastructure, clear governance, cyber resilience, privacy controls, staff capability and strong assurance.
Without those foundations, AI will struggle to scale. Worse, it may create more risk than value.
PSN polling from recent ANZ events shows the same pattern. At GIW Aotearoa 2026, 53% of Innovation Showcase respondents said they were still early stage, with some pilots and limited controls. Only 1% described their AI capability as embedded.
That is not a reason to slow down. It is a reason to build properly.
Vendors need to respect how government buys
AI does not create a shortcut around procurement.
New South Wales made this point clearly. The Office for AI is not a sales channel. Vendors still need to use the right procurement pathways, work through agency relationships and understand who owns the problem, the budget and the decision.
That is not a barrier to innovation. It is how government protects fairness, transparency and public value.
The best vendors understand this. They do not cold-pitch senior AI leaders with generic offers. They study the strategy, understand the agency context and bring useful evidence at the right time.
Reuse matters more than duplication
Government does not need more disconnected platforms.
Western Australia gave a practical example: where a solution needs a citizen interface, vendors should default to existing digital identity and authentication services rather than creating new login systems.
That principle applies broadly. Vendors should fit into existing government architecture, reuse shared platforms and reduce complexity.
Every new system creates long-term cost. It needs security review, integration, support, training and governance. Good vendors understand that and design around it.
The practical use cases are strong
The most valuable AI use cases are often simple and operational.
They include customer service support, document summarisation, internal knowledge search, reporting, compliance, case management, workflow automation, fraud detection and staff productivity.
These are not abstract ideas. They are real problems agencies deal with every day.
Government does not need every AI project to be large or high profile. It needs useful projects that save time, improve quality and make services easier to access.
The message for vendors
For vendors, the message is simple: do not sell AI. Solve a government problem. That means:
- understand the strategy before pitching
- use the right procurement channels
- show practical use cases
- explain data and governance clearly
- fit into existing architecture
- support staff adoption
- bring evidence, not hype
- respect the role of agencies and central teams
Government is open to good ideas. But buyers are becoming more disciplined. They can tell the difference between a serious public sector partner and a vendor chasing the AI trend.
The real opportunity
AI will change government, but probably not through big promises. It will change government through practical improvements: faster processes, better information, safer decisions, easier services and more confident staff.
That requires government and industry to meet in the middle.
Government needs to be clear about its priorities and standards. Vendors need to understand public sector delivery and focus on outcomes that matter to citizens.
The opportunity is not to make government look more innovative. The opportunity is to make government work better.
Want to learn more? Join our global Data & AI Community here.
Help your peers
Share what you've learned with fellow public servants