Sunburnt AI
ActiveAI Consulting & Implementation · 1-2 employees · ABN 61690890
Certification ID: RAIA-2026-0014
2026 cohort: certified ahead of the Australian Standards for AI
sunburntai.com.auTrust seal
Verified live against the certification register
Verifying against the register…In plain English
The AI Transparency Card is the one-screen version of this entry: what Sunburnt AI uses AI for, whether it makes decisions about you, who oversees it, and how to question a decision. The assurance pack is the same record arranged for a supplier questionnaire.
For AI agents
This certification is machine-verifiable. Any AI agent, search engine or procurement system can confirm it in milliseconds, no account required.
RAIA-2026-0014Responsible AI profile
Self-declared AI practices, submitted as part of this business's certification.
AI usage
How they use AI
Sunburnt AI is a Brisbane-based AI consulting and implementation firm that designs, builds and manages AI agents and workflow automations for Australian professional services businesses, including our Sunny AIOS platform for law, accounting, advocacy and financial advice firms. Internally, we use generative AI tools for research, drafting, sales operations and software development, with human review of all client-facing outputs.
AI tools in use
Pre-built models & their purpose
Agentic AI systems (Sunny AIOS) built on commercial large language models (e.g. Anthropic Claude, OpenAI GPT) for document processing, compliance auditing and workflow orchestration; AI voice agents (VAPI, ElevenLabs) for client phone enquiries; AI coding assistants for software development.
AI use & risk profile
Other risk areas
We build AI agents for clients in regulated sectors (legal, accounting, financial advice, NDIS/DVA advocacy and education compliance). These agents support document processing and compliance checks, with final decisions made by qualified professionals. Each deployment receives a use-case risk assessment before go-live.
Risk & safety
How they identify and mitigate AI risk
Every client engagement begins with a risk and use-case assessment led by our AI governance lead, covering data sensitivity, privacy obligations, accuracy and human oversight requirements. Our agents are hosted in Australia, operate read-only by default and log every action, and outputs that affect clients require human review before release. Risks are recorded in a project risk register and reviewed at each delivery milestone.
Fairness audits & human-in-the-loop review
Frequency & method
Human-in-the-loop review is built into every client agent: outputs that affect a person or a client decision are approved by a qualified staff member before release. Agent outputs are sampled and reviewed for accuracy and fairness before go-live and quarterly thereafter, overseen by Kai Baldwin.
Ability to contest or override AI decisions
Example
Our agents recommend rather than decide, and authorised staff can override or reverse any agent output. Clients and affected individuals can raise a concern with their account contact or by email, and each case is reviewed by a person, logged and escalated to Kai Baldwin where required.
Ethics & governance
Follows ethical AI frameworks
Frameworks followed
Other frameworks
Australian Government Guidance for AI Adoption (formerly the Voluntary AI Safety Standard) and the Australian Privacy Principles under the Privacy Act 1988.
Maintains AI documentation & oversight
Responsible for oversight
Kai Baldwin (Co-founder, AI Governance) holds overall accountability for AI oversight, supported by Archie Swinyard (AI Governance & Compliance Advisor). Each client build has a design document, test records, risk register and change log held in our shared project workspace, and agent activity logs are retained in our Agent Management Layer.
Privacy & fairness
Privacy & data security (Privacy Act 1988)
Methods
Other methods
Client data is hosted in Australia, AI agents run with read-only, least-privilege access by default, all agent actions are logged for audit, and we use enterprise model agreements that exclude client data from model training.
Prevents AI bias & discrimination
Methods
Other methods
We do not train our own foundation models. We select providers with published bias and safety testing, design our agents so they do not make automated decisions about individuals based on protected attributes, test outputs against varied scenarios before go-live, and keep a human in the loop for any decision affecting a person.
Sustainability & society
Societal benefit & responsible scaling
Our work focuses on giving Australian small and mid-sized professional services firms safe, practical access to AI, reducing administrative load so practitioners can spend more time with the people they serve, including NDIS participants, veterans and students with disability. As we scale, Kai Baldwin will continue to lead AI governance, supported by our AI Governance & Compliance Advisor, with a formal AI governance policy, client-level risk assessments, annual policy reviews and alignment with Australian Government Guidance for AI Adoption and Privacy Act reforms.
Certification history
Certified for less than a month. Everything below is on the public record. Purple entries are verified by Responsible AI Australia; amber entries are declared by the business.
17 November 2026
Certified at the Govern tier
Assessed by Responsible AI Australia against Australia's AI Ethics Principles.
Run a business that uses AI? See how certification works.
