Appendix B: Relationship with other frameworks
B.1 Policy frameworks
For agencies adopting AI systems, the path from policy intent to working architecture is rarely straightforward. The existing landscape, which spans strategic frameworks, legislation, ethics principles and technical standards, is rich but fragmented, leaving agencies to independently interpret and reconcile obligations. AIRA serves as a dedicated translation layer within this landscape. It bridges whole‑of‑government ambition and agency delivery by turning complex, overlapping requirements into a coherent set of reusable architectural patterns, components and expectations that agencies can readily adopt and adapt. Operational experience from agencies provides a feedback loop into AIRA. Lessons about what works, what does not work and why can be generalised across agency contexts and can used to validate, update or create reusable architectural patterns and guidance.
B.1.1 Australian and global frameworks
The following frameworks set the government's ambition and obligations for AI:
- Strategic direction is established by the Australia's AI Ethics Principles, the APS Data Ethics Framework, the AI Plan for the APS and the Data and Digital Government Strategy, which together set investment priorities and the ethical foundation for AI in government
- Legislative and mandatory obligations, include but not limited to, the Privacy Act 1988, the Archives Act 1983 and the Protective Security Policy Framework (PSPF) 2026, which set the legal boundaries within which all AI systems must operate
- Global frameworks, such as international standards, industry best practices and comparable government reference architectures from Singapore, Estonia, UK, Denmark, Canada and the European Union, inform our approach to AI
- Agency-level obligations for the use of AI are set by the Policy for the Responsible Use of AI in Government, the Digital Experience Policy and the Cloud Policy
- System-level requirements for AI-enabled solutions are established by the AI technical standard, the Digital Service Standard, the Digital Inclusion Standard, the Digital Access Standard and the Information Security Manual (ISM).
B.1.2 Australian Government Architecture
This layer enables digital coherence and value for money across government.
The Australian Government Architecture (AGA) Domain and Capability Model identify key capabilities for government digital transformation, offering a consistent vocabulary for describing government’s digital functions and supporting technologies. In this model, artificial intelligence is an enabling domain for diverse business outcomes.
The Digital Transformation Agency drives the uptake of AIRA through AGA. It provides agencies with architectural guidance aligned with whole-of-government AI policies and standards. AIRA fosters consistency and comparability across agencies while allowing flexibility in design.
As a framework, AIRA defines core components for AI systems throughout their lifecycle, highlighting standard architecture elements to promote interoperability, governance and responsible AI use.
Following AGA principles, AIRA does not mandate any products, platforms or solution designs. Each agency is tasked with creating its own solution architecture, choosing technologies and implementation methods that fit their specific business needs, environment and stage of development. These architectures should be in line with AIRA, using its shared structure, language and guidelines, while also meeting government-wide requirement for responsible AI use.
B.1.3 Australian Government agencies
Agencies will use and benefit from AIRA differently depending on their size, delivery model and in-house capability. Small agencies can use it to reduce delivery and assurance risk where specialist AI and architecture capability is limited, by relying on common patterns and leveraging shared services. Medium agencies can use it to standardise approaches across teams and programs, improving consistency of implementation, evidence and integration. Large agencies can use it to better manage complexity introduced by rapidly evolving technology, remain responsive to growing technical needs for citizen services and deliver use cases at scale through modular and reusable components, and clear interoperability expectations.
B.2 Industry architecture frameworks
Enterprise architecture frameworks help agencies organise architecture work, govern decision-making and connect policy design to outcomes, strategy, business capabilities, information, applications and technology.
AIRA is a reference architecture for government AI, at the class-of-systems tier of The Open Group Architecture Framework (TOGAF) enterprise continuum. It can be used in established enterprise architecture frameworks such as TOGAF and classification frameworks such as Zachman. It can then be instantiated into solution architectures for specific agency systems. Solution architecture remains the point at which agency-specific design decisions are made for a particular system.
B.3 Operational frameworks
Operational frameworks bridge architecture and delivery. They describe how teams organise, build, run and improve digital capabilities. They provide a common language and structure for agencies for end-to-end delivery and continuous improvement practices that automate, optimise and integrate formerly siloed processes such as development, IT operations, quality assurance and security.
Over time, many XOps frameworks have emerged, such as DevOps, FinOps and AIOps. Each addresses a different operational need, may use overlapping tools and may introduce governance gaps at the boundaries between disciplines.
AIRA can help agencies position these frameworks against architectural layers, identify likely overlaps and gaps, and consider where cross-cutting concerns need to be handled consistently. Figure 4 illustrates a mapping of operational frameworks with AIRA.
AIOps is distinct from previous frameworks. It is a capability that uses or applies AI to automate, streamline and optimise IT operations. Because AIOps is itself an AI-enabled solution, AIRA can be used to architect it.
AIRA does not require any specific operational framework. Agencies are encouraged to select and implement operational frameworks appropriate to their AI system portfolio, maturity and resourcing constraints.
