New guidance to support AI project success

As government agencies increasingly adopt and use artificial intelligence (AI) to enhance service delivery and operational efficiency, many encounter similar challenges when progressing small-scale experiments into sustainable, enterprise-ready solutions.

Building on the Policy for the responsible use of AI and the Technical standard for government’s use of AI, the DTA’s Guidance for AI proof-of-concept to scale strengthens how agencies design and deliver AI initiatives. It sets clear expectations for how proof-of-concepts (PoCs) should be developed, helping agencies plan them around defined business outcomes and principles.  

Why AI proof of concepts are crucial

“We’ve heard from many agencies that getting AI proof‑of‑concepts off the ground can be hard,” explains Lucy Poole, Deputy Chief Executive Officer, Strategy, Planning and Performance at the DTA.  

“What our experience shows is that the real value comes when trials are planned with their end use in mind. That means being clear from day one about who will use it, where it will sit in a real workflow, what decisions it will inform, and what evidence you’ll need to justify scaling it.”

An AI PoC is a focused, small-scale experiment designed to demonstrate technical feasibility and potential business value of an AI use case. As a vital first step, PoCs help agencies test ideas early and build the confidence needed to determine whether an AI solution is suitable for enterprise-wide scaling, without committing significant time or resources.  

Agencies that successfully transition AI solutions from PoC to sustained operational use consistent characteristics: coordinated system level planning, early integration of governance requirements and evidence-based decision making.  

Without these foundations in place from the outset, even high performing PoCs are unlikely to achieve enduring impact for government.

The 8 principles that guide success

The Guidance for AI PoC to Scale outlines eight guiding principles that capture the essential factors for progressing AI solutions beyond experimentation:

1. Strong foundations – core capabilities (data, talent, tools and processes) are in place and actively maintained to support AI at scale.  
2. Enterprise-ready design and infrastructure – solutions are built with scalability, interoperability and operational resilience in mind.
3. Robust governance and trust frameworks – clear policies, risk controls, accountability and ethical safeguards guide responsible AI use and ensure compliance with the AI in government policy.
4. Cross-functional collaboration and accountability – alignment across technical, business, legal and operational teams ensures shared ownership, accountability and long-term sustainability of AI initiatives.
5. Strategic alignment and measurable outcomes – AI initiatives are tied to business priorities, with defined success metrics and baselines, systematic evaluation methods and pathways to value.
6. Culture of responsible innovation and business value – a culture that embraces continuous learning, ethical innovation and business impact encourages experimentation.
7. AI literacy at all levels – AI literacy is promoted across the agency (from staff to leadership), ensuring a shared understanding of AI concepts, opportunities, risks and responsible practices.
8. Right technology for the right problem – technology choices are driven by the business problem to be solved – not by novelty or trend.

“By considering governance from the outset, how it will fit with existing systems, and what success looks like, you can turn what works into something that lasts and delivers real benefits for government and the people it serves.”

Applying these principles from the outset helps set AI initiatives up to be responsibly scaled and embedded into day-to-day operations. They have been embedded in practical tools, including an evaluation guidance and AI readiness checklist, to support agencies at every stage of the AI lifecycle.

“It also means that when a PoC doesn’t deliver the expected results, agencies can clearly understand why. From there, they can determine how to make better and more informed decisions, rather than discarding good ideas because the right foundations weren’t in place.”

This guidance helps agencies move promising AI experiments into lasting capability by embedding governance, scalability and strategic alignment from the beginning. This increases the likelihood that valuable AI initiatives progress into sustainable, enterprise-ready solutions.

The Guidance for AI Proof‑of‑Concept to Scale is now available at digital.gov.au.