Field notes.
Long‑form essays on AI under AFSL. What we ship, what we refuse, and what the cloud vendors will not tell you.
- AI Governance28 September 2026 · 5 min read
What APRA's April 2026 Letter on AI Asks of Regulated Entities
APRA's 30 April 2026 letter found AI assurance lagging and set out what it expects: an inventory, human involvement, and a supply chain you can see.
- Compliance28 September 2026 · 6 min read
CPS 230 Offshoring: Notify APRA Before, Not After
CPS 230 splits two notices that are easy to confuse. Offshoring must be notified before you sign, and what counts is where the service is performed.
- Data Sovereignty28 September 2026 · 6 min read
Does Australian Law Require Financial Data to Stay Onshore?
No general rule keeps financial services data in Australia. The Privacy Act, CPS 230 and the licensee obligations ask for something harder: control you can prove.
- Data Sovereignty28 September 2026 · 5 min read
Is OpenAI's Australian Data Residency Enough for a Fund Manager?
OpenAI now stores eligible business data in Australia. That answers where data rests, not where it is processed, and the obligations a fund manager holds still apply.
- Advice25 September 2026 · 7 min read
How to Evaluate SoA Automation: Five Questions Worth Asking Any Vendor
How to evaluate SoA automation before you buy: five questions about templates, deterministic checks, data location and what the software refuses to do.
- Assurance24 September 2026 · 5 min read
Checking a Number Against a Bag of Numbers Is Not Checking
Verifying figures in an AI-drafted Statement of Advice: why checking each number against a bag of numbers missed every substitution, and what replaced it.
- Advice17 September 2026 · 5 min read
The Three Ways a Language Model Breaks Inside a Statement of Advice
Three ways a language model fails inside a Statement of Advice, and why a naive refusal detector ruins a document that quotes the client.
- Advice17 September 2026 · 6 min read
What Has to Be True Before a Licensee Accepts an AI-Drafted Statement of Advice
What a licensee should require before accepting an AI-drafted Statement of Advice: which checks are deterministic, and what stays the adviser’s judgement.
- Advice17 September 2026 · 5 min read
Records of Advice Are Where the Time Actually Goes
Record of Advice automation is where the hours repeat. Why an advice practice should price the ongoing review cycle rather than the new-client SoA.
- Advice17 September 2026 · 6 min read
Your Template, Not Ours: How SoA Automation Works With a Firm's Own Document
How SoA automation works with a firm’s own Statement of Advice template: three kinds of bracket, and only one of them ever reaches a language model.
- Architecture16 September 2026 · 5 min read
A Confidence Score Is Only Useful if the System May Act on It
Most retrieval pipelines produce a confidence number and answer regardless. What changed the refusal rate in our March 2026 benchmark was letting it stop.
- AI Governance23 July 2026 · 12 min read
Beyond Reporting: Is Enterprise AI Set to Become the Enterprise Reasoning Layer?
Most enterprise AI just bolts a chat box onto existing software. The more useful question is how enterprise software participates in enterprise reasoning: a governed layer that combines a language model with controlled retrieval, deterministic rules and human approval to assemble evidence across systems and produce a defensible Total Portfolio View.
- AI Governance20 July 2026 · 6 min read
Why General-Purpose LLMs Are the Wrong Tool for Regulated Financial Advice
Financial firms are pasting client data into Copilot, ChatGPT and Claude to draft Advice Statements and DDQs, workflows these tools were never designed to perform. Why regulated advice needs data custody, auditability, deterministic source-grounded outputs, and clear accountability, none of which a general-purpose model can add after the fact.
- AI Governance17 July 2026 · 5 min read
Retrieval Laundering: The Next Governance Challenge for Enterprise AI
RAG grounds AI in your own knowledge base, but it introduces a new governance risk: Retrieval Laundering, where genuine but outdated, conflicting, or inappropriate documents are treated as authoritative. Why it is emerging as the defining governance challenge for enterprise AI, and how to mitigate it.
- Financial Advice3 July 2026 · 6 min read
Australia's Advice Capacity Crisis: Another Perfect Storm
Compliance overheads are up and adviser numbers are down while retirement advice grows more complex. Here is why Australia faces an advice capacity crisis and how automation expands adviser output.
- Compliance25 March 2026 · 9 min read
CPS 234 Compliance Checklist for AI Tools in Financial Services
APRA's CPS 234 applies to every AI tool processing client data. This checklist covers what Australian financial services firms must verify before adopting AI: information security, third-party risk, access controls, and incident reporting.
- Data Sovereignty25 March 2026 · 8 min read
Why Australian Fund Managers Can't Use ChatGPT for Client Documents
ChatGPT sends data to US servers, trains on your inputs by default, and offers no accuracy warranty. For Australian fund managers handling DDQs, investor reports, and compliance documents, that is a regulatory problem.
- Financial Advice13 March 2026 · 7 min read
SOA Automation in Australia: How Financial Advisers Are Cutting Prep Time by 60%
Six to eight hours per Statement of Advice is the figure the industry quotes. What automation actually removes from that, what it cannot, and where the adviser's judgement stays.
- Fund Management12 March 2026 · 7 min read
DDQ Automation for Fund Managers: From 3 Days to About an Hour
Due diligence questionnaires eat senior time at every Australian fund manager. How source-cited AI, deployed in the manager's own environment, drafts DDQ responses.
- Superannuation11 March 2026 · 8 min read
APRA Compliance Automation for Super Funds: AI That Runs Inside Your Infrastructure
Australian super funds face growing APRA obligations. How AI deployed inside the fund's own infrastructure supports CPS 234, member services and trustee documentation.
- Insurance10 March 2026 · 7 min read
AI Claims Processing for Australian Insurers: On-Premise Automation Under APRA's Standards
Australian insurance companies face mounting claims volumes and APRA compliance pressure. On-premise AI automates claims document analysis, policy interpretation, and regulatory reporting.
- News10 March 2026 · 2 min read
BackPro AI Now Available on the Microsoft Azure Marketplace
BackPro AI's compliance-native document automation platform is now listed on the Microsoft Azure Marketplace, enabling regulated organisations to deploy on-premise AI directly from their Azure environment.
- Data Sovereignty9 March 2026 · 8 min read
Why On-Premise AI Is Non-Negotiable for Australian Financial Services
Cloud AI raises hard questions for Australian financial services firms. Running AI in infrastructure you control makes the Privacy Act, ASIC and APRA answers easier to give.
- News14 December 2025 · 2 min read
BackPro AI Wins Inaugural MLC Financial Clarity Prize at Genesis Finale
BackPro AI was awarded the inaugural MLC Financial Clarity Prize at the University of Sydney's Genesis Accelerator Finale, presented by Insignia Financial for excellence in financial modelling.
Take this with you · eBook · PDF, 11 pages
The AI-Native Compliance Advantage
The questions advisers and licensees ask about AI in advice documents, from "will it make things up?" to "who is responsible?", answered with the evidence.
Sent to your inbox. No call, and nothing else unless you ask.