How AI adoption can drive better government services

Atturra

By Chris Seysener, General Manager – AI Strategy and Transformation, Atturra
Friday, 20 November, 2026


How AI adoption can drive better government services

Artificial intelligence has rapidly established itself across Australia’s public sector, where employees are using AI-powered tools to draft documents and complete administrative tasks that once consumed hours of each working week.

Adoption continues to accelerate, but government executives now face a harder question: is this investment delivering better public services? Saving an employee an hour each day might represent a welcome productivity improvement, but it does not automatically translate into a better experience for citizens.

Evidence suggests AI adoption across NSW Government is growing, while oversight of its use and costs remains incomplete. The NSW Auditor-General’s ‘Internal controls and governance 2025: Procurement and technology’ report found that agencies reported 357 different AI tools, either in pilot or fully implemented, across 21 of the 26 large NSW public sector agencies included in the report. The Auditor-General’s 2026 internal controls and governance report examined AI governance at 10 selected agencies with the highest AI use or greatest potential benefit from AI, and found that most did not centrally track AI costs or set budgets for AI solutions. Without reliable cost and benefit data, government leaders cannot determine whether growing AI adoption is translating into better services for citizens.

A Commonwealth trial found some employees saved about an hour each day on drafting, searching and summarising. However, a separate Treasury evaluation found no clear evidence that the technology had improved work outcomes.

For government leaders, the message is becoming increasingly clear. The next phase of AI adoption must focus less on the technology and more on the outcomes it can deliver.

Moving beyond individual productivity

A useful starting point is to recognise that everyday AI adoption and genuine service transformation are two different, though closely related, activities.

Everyday AI involves equipping public servants with tools that help them work more efficiently. It includes improving AI literacy and encouraging employees to identify opportunities to reduce repetitive work.

Genuine service transformation requires a broader perspective. Rather than asking how AI can help an individual employee complete a task more quickly, executives should examine how an entire service can be redesigned to deliver a better outcome.

Consider a government department responsible for assessing development applications. AI might allow an employee to summarise supporting documents in minutes rather than hours. However, if the application process doesn’t capture the required information upfront, it must still pass through multiple approval stages and undergo repeated manual checks, and the overall improvement may be marginal.

The bigger opportunity is to understand why the process takes so long in determining which activities can be eliminated, simplified or automated. This requires agencies to begin with their service mission, rather than the capabilities of a particular AI product.

Fix the process before choosing the technology

One of the biggest risks facing government is the temptation to apply sophisticated technology to existing inefficient processes. Automating a poorly designed workflow may simply allow an agency to complete unnecessary activities more quickly: a better approach is to examine the work itself and redesign the process before choosing the technology.

NSW development applications show why agencies should examine the process before choosing the technology. Councils assess about 85% of NSW’s housing development applications, and almost 60,000 development applications are lodged across NSW each year. In one NSW housing program, 30% of the applications examined required further information, and applicants took an average of 42 additional days to respond. The NSW Government estimated that avoiding requests for further information could save 6300 days across that group alone.

The first opportunity is therefore not necessarily a complex AI system: clearer requirements and rules-based checks can help establish whether mandatory information and documents are present before an application enters assessment. AI becomes more useful at the next stage, where plans and supporting reports need to be read and checked for missing or inconsistent information. An officer should confirm those findings and retain responsibility for any statutory assessment or determination.

Applying AI to every stage would add cost and risk without necessarily improving the service. The better approach is to redesign the application journey, apply the simplest technology appropriate to each step, and measure whether it reduces requests for information and days to decision.

Building the foundations for sustainable change

Agencies moving beyond experimentation need strong operational foundations. This does not always mean fixing every system or data source before starting: it means understanding the information, integrations and controls required for the service being redesigned.

A large NSW local government department recently implemented changes to improve its data governance, to understand what it held ahead of scaled AI adoption. The council had fragmented legacy file shares, little visibility, unsupported infrastructure, and mounting compliance and cyber risk. The scan involved 11.5 million unknown files, with an objective of removing redundant and obsolete data, while enriching the metadata to enable AI. Richer metadata helps AI find the right content and lets policy-based controls manage personal information risk.

Improving services requires more than adding an AI tool. Information must move reliably between systems, documents need consistent classification, access must be controlled and AI-generated outputs must be traceable. The opportunity is to establish those foundations around a defined workflow, test whether they reduce rework and assessment time, and scale only when the evidence supports it.

The resulting information environment supported enterprise search, improved control of personal information, and AI-assisted responses grounded in retrieved documents. The underlying data improvements created value across multiple activities rather than supporting a single AI application.

Turning promising pilots into production services

The challenge for government executives is moving successful AI experiments into sustainable operational environments. A demonstration may show that an AI tool can complete a task, but production deployment requires evidence that it can perform reliably under real-world conditions. Before approving wider implementation, agencies should establish clear performance baselines and assess whether the proposed solution improves outcomes that matter to service users. These measures might include processing times, application error rates and the quality of decisions.

The most effective government AI strategies will connect everyday adoption with broader service transformation. Employees using AI in their daily work are well placed to identify recurring problems and opportunities for improvement. Agency leaders can use these insights to validate the department’s strategy and identify services that warrant more substantial redesign, while successful transformation initiatives can generate reusable data, processes and capabilities for other teams.

The immediate priority should be to identify a service where better outcomes would make a tangible difference, establish a baseline, and determine the smallest practical intervention capable of improving performance. That intervention may involve AI, but it should not be assumed from the outset.

The success of government AI investment will not be determined by how many employees use the technology, how many pilots are launched, or how sophisticated the systems become; it will be judged by whether citizens receive faster, more reliable and more accessible services.

Government has already demonstrated its willingness to adopt AI, and the challenge now is to prove that adoption can translate into meaningful public value.

Top image credit: iStock.com/Stock photo and footage

Related Articles

AI could help government solve its most stubborn technology problems

AI gives us a new opportunity to tackle some of the most longstanding challenges in government...

How AI is driving a fresh wave of transformation in local government

Councils are taking a more strategic approach to their investments in ERP software, with AI...

The Budget backs local government: now tech needs to keep pace

How councils can turn this year's budget funding into lasting value for their communities.


  • All content Copyright © 2026 Westwick-Farrow Pty Ltd