What Your Clients Are Asking AI Before They Ask You

In Summary

• Almost half of pre-retirees have used a generative AI tool, and advised clients use AI more than unadvised ones, most often to explore options and prepare for adviser meetings.
• AI has closed very little of the gap that matters: 41 per cent of pre-retirees understand what they will need to fund retirement, against 68 per cent of advised pre-retirees.
• Compliance thinking is moving towards AI as a formal control inside governance frameworks, which makes how your paraplanning partner uses AI a question worth asking directly.

Two articles published in Professional Planner this month say something useful when read together. One looks at how pre-retirees are using generative AI. The other argues that responsible AI belongs inside every compliance framework.

Between them they describe the same shift from two directions. AI has arrived in the advice relationship, at the client end and at the governance end.

The Client Has Already Asked the Chatbot

The CoreData/Conexus Financial Best Possible Retirement study found that 34 per cent of pre-retirees and retirees have used a generative AI tool for general enquiries. Among pre-retirees the figure is 49.6 per cent, against 20.7 per cent of retirees.

Usage falls steadily with age, from 55.9 per cent of 45 to 49 year olds down to 13.5 per cent of those aged 70 and over.

The more interesting finding is who is using it. Advised clients are more likely to have used a generative AI tool than unadvised clients, 38.1 per cent against 32.9 per cent.

The industry has tended to assume clients turn to AI because they cannot afford or cannot access an adviser. The data points somewhere else. Advised clients’ most common reason for using AI was wanting to explore their options, at 41.5 per cent, followed by convenience and wanting information outside business hours.

Difficulty getting hold of someone at their adviser’s practice came last, cited by just 7.5 per cent.

CoreData research consultant Alana Devitt notes that almost a quarter of the people using AI for retirement questions are using it to draft what they will ask their adviser next. She describes that behaviour as preparation.

What This Means in the Meeting Room

Clients arriving with AI-generated questions is now normal, and it will become more so as the cohort using these tools moves into retirement. Some of what they bring will be sound. Some will be generic, out of date, or drawn from overseas rules that have no application in Australia.

Treating that as an intrusion wastes an opportunity. A client who has explored their options before the meeting is an engaged client, and correcting a misconception is easier work than creating interest from scratch.

The practical response is to ask. What did you look up? What did it tell you? That makes the chatbot a starting point for the conversation, which is roughly what your clients are already using it for.

It also helps to know where AI still falls short. The same study found only 41 per cent of pre-retirees understand how much they will need to fund retirement. Among advised pre-retirees, that rises to 68 per cent. The gap AI has failed to close is the one that matters most.

Compliance Frameworks Are Catching Up

The second article, written by 3Lines co-founder Julia Vojkovic, makes the case from inside the compliance function.

Her observation from years of advice file reviews is that most adverse findings turn out to be process gaps rather than misconduct: an overlooked disclosure, an outdated template, a missing supporting document, a strategy that was not explained clearly enough. These come from time pressure and regulatory complexity, and they happen to competent people.

Her argument is that AI is well suited to catching this category of problem, because it performs repetitive, structured checks with a consistency that reviewers working to deadlines struggle to sustain across hundreds of files.

The caveat matters as much as the argument. AI used in governance needs governance of its own. Businesses should understand what a tool is designed to do, where its limits sit, and where professional judgement has to prevail.

Her view is that the bigger risk has now shifted from organisations adopting AI to organisations failing to adopt it.

The Question to Ask Your Paraplanner

Both developments land in the same place for advisers: the advice documentation process.

If your paraplanning partner uses AI, you are entitled to know how. Which tasks it touches, what human review sits over its output, where client data goes, and how it is documented if a file is ever reviewed.

If your paraplanning partner avoids AI entirely, that is a legitimate position too, provided it is a considered decision with reasoning behind it.

Either way, the answer should be specific. A vague reassurance about human oversight is exactly the governance gap Vojkovic describes, and it tends to surface at the worst possible moment.

Bottom Line

Clients are using AI to prepare for conversations with you. Compliance professionals are building AI into the frameworks that review your files. Both are happening whether or not your practice has taken a position on it.

The advisers who handle this well will be the ones who treat AI as another process to govern: understood, supervised, documented, and pointed at the tasks it genuinely does well.

At Mutual Plans, we are happy to have that conversation in as much detail as you want. It is the sort of question every adviser should be putting to their paraplanning partner right now.

Frequently Asked Questions

Are clients using AI instead of seeing an adviser?

The research points the other way. Advised clients are more likely to have used a generative AI tool than unadvised clients, 38.1 per cent against 32.9 per cent, and difficulty accessing their adviser’s practice was the least cited reason for turning to AI, at 7.5 per cent.

What are clients actually using AI for?

Exploring options was the most common reason among advised clients, at 41.5 per cent, followed by convenience and wanting information outside business hours. Retirees most commonly ask AI how long their money will last and what government support is available to them.

Should advisers discourage clients from using AI?

Discouragement is unlikely to work, and it removes useful information from the meeting. Asking what the client looked up and what they were told gives the adviser a chance to correct misconceptions and build on the client’s own thinking.

Why does AI belong inside a compliance framework?

Because it is already being used informally in most advice businesses. Bringing it inside the framework means defining what each tool does, where its limits sit, what human review applies, and how its use is documented. That treats AI the same way as peer review, quality assurance and file monitoring.

What are process gaps and why do they matter here?

They are the documentation shortfalls that make up most adverse file review findings: an overlooked disclosure, an outdated template, a missing supporting document. They come from time pressure rather than misconduct, and they are the category of problem structured checking is best suited to catch.

What should I ask my paraplanning partner about AI?

Which tasks AI touches, what human review sits over the output, where client data is stored and transmitted, whether client information is entered into third-party tools, and how the process is documented if a file is reviewed. The answers should be specific.

Glossary

Advice file review. A structured review of client files against licensee and regulatory requirements, testing documentation, disclosure and the basis for recommendations.

AFSL. Australian Financial Services Licence. The licence a business needs to provide financial services, carrying obligations for the supervision and monitoring of its representatives.

Data governance. The rules covering how client information is stored, transmitted, accessed and retained, including whether it may be entered into third-party AI tools.

Generative AI. Software that produces new text, images or other content in response to prompts. ChatGPT, Copilot, Gemini and Claude are the tools consumers name most often.

Hallucination. An AI output that reads as fluent and confident while being factually wrong. A live risk wherever advice content touches legislation, thresholds or product terms.

Human in the loop. A design principle requiring a qualified person to review and approve AI output before it is used or relied on.

Large language model (LLM). The category of AI system behind most generative tools, trained on large volumes of text to predict and produce language.

Process gap. A documentation or procedural shortfall in an advice file, such as a missing disclosure or an outdated template, distinct from deliberate misconduct.

Quality assurance. Systematic checking of work against a defined standard, applied before or after delivery.

RegTech. Technology built specifically to support regulatory compliance, monitoring and reporting.

Responsible AI. The practice of deploying AI within defined governance settings covering purpose, limitations, oversight, data handling and accountability.