The terrain · standards & obligations
The standards and obligations we build for.
AI now comes with duties. Australian policy, the regulators and the international standards all assume an organisational layer exists around AI: accountable ownership, decision gates, human oversight, monitoring and evidence. Most organisations have not built it. That layer is what we build, and this is the terrain it maps to.
The home terrain
Australian policy and regulation.
Government AI policy · DTA v2.0
The DTA Policy for the responsible use of AI, v2.0. Agencies must run AI use-case impact assessments for in-scope systems by 15 December 2026, with accountable officials, use-case registers and ongoing monitoring. We help agencies stand up that operating model and facilitate the assessments.
Voluntary AI Safety Standard · National AI Centre
Ten guardrails covering accountability, risk, testing, human oversight, transparency and records. We read how those guardrails operate in practice, not just whether they are written down.
APRA CPS 230 · Operational risk management
AI systems and agents are operational-risk surfaces and service providers. We design the controls, testing, monitoring and orderly-exit conditions AI introduces. Regulated engagements run with an APRA-experienced partner.
ASIC governance expectations · REP 798
ASIC has named the gap between AI adoption and licensee governance. We build the operating evidence that shows governance works in practice.
Privacy Act & OAIC guidance
Privacy-by-design and impact assessments before adoption, per OAIC guidance on commercially available AI. We build the readiness gates and procurement controls that make bought AI safe to adopt.
Safe Work Australia psychosocial duties · ISO 45003
Workload, role clarity, job control and poorly managed change are psychosocial risks, and AI-driven work redesign is one. Our work-design and adaptive-experience methods are built around these exact dimensions.
The international frame
International standards.
ISO/IEC 42001 · AI management systems
We build the operating substance a management system needs, and prepare organisations for conformity assessment. We do not certify; certification is performed by accredited bodies.
ISO/IEC 42005 · AI system impact assessment
We run impact assessments consistent with this guidance and the Commonwealth policy.
ISO/IEC 23894 · AI risk management
Our readiness gates, decision rights and monitoring design align to this risk-process guidance.
ISO/IEC 42006 · Audit & certification bodies
This standard formalises assessor independence. It is why we separate the people who build from the people who review, and use an external reviewer where reliance requires it.
NIST AI Risk Management Framework
Govern, map, measure, manage. Our Qualitative Coherence Indicators sit in ‘measure’; Coherence OS spans ‘govern’ and ‘manage’.
EU AI Act
For organisations with European exposure, we design the human oversight, monitoring and record-keeping the Act expects of high-risk systems.
WCAG 2.2 & ISO 9241-210 · Accessibility & human-centred design
We design AI experiences to these baselines, including for neurodivergence.
The call: 30 minutes · your obligations, your systems, our read · no deck.
Divergent Kind facilitates conditions for agency and coherence. We do not diagnose, treat, or cure.
We do not determine legal compliance, replace your accountable officials, or act as your auditor or certifier. We build and evidence the organisational layer those duties assume.
This terrain moves: ISO/IEC 42005 and ISO/IEC 42006 were both published in 2025, and Australian standards legislation is expected in 2027. Page dated 24 July 2026; reviewed quarterly. Standard and framework names are cited for identification; alignment statements describe how our method is designed, not a conformity finding, certification or legal advice. Divergent Kind Pty Ltd · ABN 30 693 755 672 · Australian SME.