Australian wealth compliance teams review advice files against the best interests duty, check that ongoing fees still rest on current client consent, assess breaches for reporting to ASIC and test source of wealth. AI agents can read every advice file and fee record and draft MLRO reporting. People decide breaches and remediation. ASIC, APRA and the OAIC supervise.
What fills a wealth compliance team's week in Australia?
Advice file review is the core. Each statement of advice must show that the adviser understood the client's circumstances, considered alternatives, and recommended what was in the client's best interests. Reviewers check the fact find, the reasoning, product comparisons and replacement product analysis, and they look harder at advice to switch superannuation, where fees and insurance can be lost.
Fees are the second strand. Clients on ongoing advice arrangements must renew their consent periodically, and fees charged without current consent, or for services not delivered, lead to remediation. Compliance tests that consent records, fee deductions and services line up.
Then there is breach reporting. Licensees must assess whether incidents are reportable to ASIC within set timeframes, and remediate affected clients. Product governance, AML checks on high-net-worth clients and privacy fill the rest.
Where can AI agents help?
- Advice file review (depth 1). The agent reads the fact find, the statement of advice and the product documents, and checks the reasoning against the firm's best interests checklist. It flags gaps: goals recorded but not addressed, a switch recommended without comparing the insurance being given up, fees not explained.
- Ongoing fee consent testing (depth 1). The agent compares each client's consent dates with fee deductions from the platform and records of services delivered, and lists fees charged after consent lapsed or without the promised review meeting.
- Source of wealth assessment (depth 2). For new high-value clients, the agent reads the declared source of wealth, compares it with supporting evidence such as a sale contract or company accounts, and drafts an assessment with gaps listed. An analyst decides.
- MLRO reporting (depth 2). The agent compiles alert volumes, aged cases and high-risk client reviews into the MLRO's monthly pack. Our guide on how to prepare MLRO management information with an AI agent sets out the steps.
The agent needs reliable access to case history. When it flags an advice file, the flag should create a case in your existing system, not a separate list. Our guide on connecting an AI agent to your case management system covers how to do that without giving the agent authority to close cases.
What stays with people?
Whether advice was in the client's best interests is a professional judgement. Two reviewers can disagree, and the conclusion affects the adviser's standing and the client's money. The agent marks gaps; a qualified reviewer concludes.
Breach reporting decisions belong to named staff. Whether an incident is significant and reportable to ASIC is a legal judgement, often with counsel, and the timeframe makes it important that the agent surfaces incidents early so people have time to decide.
Remediation decisions, such as refunding fees to every client whose consent lapsed, and any action against an adviser stay human. So does the decision to accept or decline a client on source of wealth grounds.
Which rules apply in Australia?
ASIC regulates financial advice, product disclosure and conduct. Its regime includes the best interests duty for personal advice, consent requirements for ongoing fee arrangements, design and distribution obligations for financial products, and a breach reporting regime with set timeframes. APRA supervises superannuation trustees, and its CPS 230 standard on operational risk management applies to APRA-regulated entities, including their reliance on service providers and critical operations. An agent embedded in those operations is within scope.
AML/CFT duties follow the FATF Recommendations, including customer due diligence and enhanced checks for higher-risk clients. The Privacy Act 1988, administered by the OAIC, governs client personal information, including the financial and family details found in every fact find.
Australia has no AI act. The government's voluntary AI safety standard is a sensible reference for governance, testing and human oversight. Rules here change often, so treat the agent's sources as part of your regulatory change management: when a rule changes, the checklist the agent applies must change too, and its certificate should be re-tested.
This is an explanation of the rules, not legal advice; confirm how they apply to you with counsel.
How to start in the first 90 days
Month one: begin with ongoing fee consent testing. It is rule-bound, uses platform data and has a direct link to client harm. Name the officer who owns findings and agree the sample.
Month two: replay the agent on a past period where your team already ran a fee review, including the remediation that followed. Check whether it would have found every affected client.
Month three: certify at depth 1, then start advice file review on superannuation switching advice with a senior reviewer checking every flag. Nigerian wealth managers face a different mix of risks; see our brief on compliance agents in Nigerian wealth management. The broader stance is in AI compliance agents: second line, never the last word.
Questions people ask
Can an agent decide whether advice met the best interests duty?
No. It can check the file against your checklist and point to gaps. The conclusion is a professional judgement for a qualified reviewer.
Does the agent need access to the adviser's emails?
Usually not for file review. The statement of advice, fact find and platform records are enough. Add correspondence only for specific investigations, with a recorded reason.
How often should consent testing run?
Monthly suits most licensees, because consent lapses on a rolling basis. Running it monthly catches lapsed consents before many fees are charged.