Guidance for AI proof of concept to scale

Appendix 1: AI readiness checklist for sponsors, managers and practitioners

This checklist is designed to guide agencies in maturing their AI practices. It supports confident, responsible adoption by outlining key considerations at each major stage of the AI journey – from initial proof of concept (PoC) to enterprise-scale deployment. By working through these questions, agencies can strengthen strategic alignment, reduce common risks and establish the foundations for sustainable and impactful AI innovation.

1. Before starting a proof of concept (PoC)

Strategic alignment

1. Is the PoC linked to a clear business priority or public outcome?
2. Have success criteria and measurable outcomes been defined?
3. Is there a clear understanding of expected costs and benefits?
4. Has the Benefits Management Policy been followed?
5. Have approvals been provided by the Accountable Official and the Chief AI Officer?
6. Has the proposed PoC been checked for reuse opportunities and reducing duplication via the agency's internal use case register?

Design & approach

7. Is AI the right solution? If so, is the selected type of AI fit-for-purpose?
8. Can AI realistically deliver the intended outcome?
9. Is this a novel use case or has it been done before?
10. Has product thinking and user-centric design been embedded from the outset, and is there a baseline understanding of user needs and current workflows?
11. Are stakeholders (business, IT, legal etc.) engaged early for co-design and transparency?
12. Has the business been consulted and is there a principal Business Owner involved that can ensure proposed outcomes are validated?
13. Does the approach align with the Technical standard for government's use of artificial intelligence?
14. Are there any workforce implications/considerations that need to be managed?
15. Has a decision register been established?
16. Has a lessons learned log been established?

Data & infrastructure readiness

17. Is the required data available, high-quality and appropriately governed?
18. Are integration points with existing systems well understood?
19. Are automated pipelines in place to track experiments and results?

Governance, risk & impact

20. Have ethical, privacy and security considerations been addressed? These are outlined in the AI Policy, especially AI Policy Appendix C depending on use case.
21. Is there a governance framework in place for oversight and accountability?
22. Has an impact assessment been conducted for the appropriate use case using the AI Impact Assessment Tool?

2. After PoC completion

Value assessment

1. Did the PoC meet the defined success criteria?
2. What tangible benefits were demonstrated (e.g. efficiency, cost savings, accuracy)?
3. Is there buy-in from prospective user groups?
4. Should progress be made to scale-up the AI?
5. Were the Benefits Management outcomes met?
6. Is there a clear pathway to deploy to production and adequate funding available?

Stakeholder engagement

7. Is there leadership buy-in and trust in the solution?
8. Have all privacy considerations been factored in? Considering the Office of the Australian Information Commissioner (OAIC) as a foundational stakeholder.
9. Is there business sponsorship?
10. Have results been communicated in business terms, not just technical metrics?
11. Are there any workforce impacts that need to be supported and engaged through the implementation process?

Technical feasibility

12. Have any technical debt or integration challenges been identified?
13. Can the solution operate within existing ICT constraints?

3. Before scaling (pathway to production)

Enterprise eeadiness

1. Are there clear steps for further development, validation, integration and testing (including regression testing and user testing)?
2. Has adequate funding been sought and approved to cover the pathway to production? Is there a provision for top-up funding if needed?
3. Is the solution designed for scalability, interoperability and resilience?
4. Are modular, reusable components and a composable architecture in place?
5. Is the data breach response plan still applicable for the solution being assessed?
6. Has planning commenced to support any potential workforce impacts where relevant?
7. Has the Technical standard for government's use of artificial intelligence been considered?

Operational processes

8. Are automated pipelines in place for continuous deployment and monitoring?
9. Are natural change management processes defined to handle updates smoothly?

Governance & compliance

10. Are policies, risk controls and ethical safeguards embedded for scaled use? Agencies need to evaluate against the AI Policy, especially AI Policy Appendix C dependent on use case.
11. Is there a clear ownership and accountability model across teams?
12. Has an impact assessment been conducted?

4. Pilot phase (controlled real-world testing and validation)

Real-world testing

1. Is the solution being tested in a controlled, representative environment?
2. Are diverse user groups and subject matter experts involved in testing and validating the AI solution?

Equity, inclusion and diversity

3. Are equity criteria embedded in pilot evaluation?
4. Are digital literacy supports and accessible feedback channels in place?
5. Has the Digital Experience Policy and Digital Inclusion Standard been reviewed?

Impact monitoring

6. Are feedback loops established to refine the AI solution before scaling?

Operational readiness

7. Are operational processes to manage AI trialled (e.g. support, monitoring, updates)?
8. Is the relevant workforce supported through potential changes (e.g. training and upskilling)?

5. After scaling (operationalisation)

Monitoring & adaptation

1. Are continuous monitoring and performance tracking mechanisms in place?
2. Are feedback loops for model retraining and improvement working effectively?

Change management & culture

3. Are change management processes working effectively to support adoption?
4. Is there a culture of learning, sharing and reuse across teams?

Sustainability

5. Are long-term capability-building and talent development plans in place?
6. Is governance evolving to support responsible innovation at scale?
7. Are mechanisms in place to monitor compliance with the AI in government policy?