Issue 03 · 4 August 2026
More than half will accept AI in their financial advice, on one condition. Plus the advice plumbing quietly turning into a land grab, and Model Context Protocol showing up in an Australian advice deal.
The Take
deVere Group surveyed clients in July and found 55 per cent are comfortable with AI in financial planning. The interesting part is the condition attached: only if a human adviser keeps the final say.
James Green from deVere put it better than I could. "They want it doing the work nobody enjoys, not making the calls that decide their retirement." Hand the decision over without a person behind it, he reckons, and "you've not modernised the relationship. You've probably ended it."
Here's what struck me. The industry keeps asking how much AI it can get away with, as though clients are sliding a dial between none and lots. They aren't. They've drawn their line somewhere else entirely, around accountability. Do the groundwork, all of it. Just have a person who owns the answer.
Which means the constraint isn't really a constraint. Build the way you should be building anyway, machine drafts and a human approves before anything reaches a client, and you're already on the right side of it. That approval step isn't friction in the workflow. It's the product.
The firms with a problem are the ones treating it as the last inefficiency to engineer out. Jasmine Blackham's warning describes a failure you won't see coming: clients "don't complain and drift away quietly when a firm crosses this line. They leave fast."
One caveat. This is a global adviser network surveying its own client base, not regulator data and not Australian-specific, so treat the 55 per cent as directional.
Also Worth Knowing
Same survey, second finding, and it's the operationally useful one. Attitudes diverge sharply by age. Clients around 54 will experiment with AI happily. By about 65, scepticism is the default. Strategy consultant Jasmine Blackham: "A fifty-something client will experiment with it happily. A client in their mid-sixties, often with far more at stake, wants a person they trust between them and the tech, not an app standing in for one."
Read that next to the CoreData numbers from the last issue, where AI use ran at 49.6 per cent among pre-retirees and 20.7 per cent among retirees, and a practical instruction falls out. Your disclosure about how you use AI shouldn't be one paragraph applied to every client. The accumulation-phase client wants to hear you're using good tools. The client three years into drawdown wants to hear that a person checked it. Same practice, same technology, two genuinely different conversations. Most practices are having neither.
Padua has signed AUSIEX as exclusive broking partner on its Advice Portal, going after what chief executive Matthew Esler calls the sector's "last mile", the gap between giving the advice and actually implementing it. Buried in the detail is the bit worth your attention: Esler named API and Model Context Protocol integration as the portal's core feature, saying those integrations "reduce manual handling, improve implementation accuracy and support straight-through processing". Padua expects $27 billion in active money recommendations through the ecosystem in FY2027, up from $17 billion in FY26.
MCP is a young open standard for letting AI systems talk to tools and data in a consistent way. Seeing it named in an Australian advice-tech announcement, rather than in a developer conference talk, is a genuine marker. What it signals is that the plumbing layer is being built to be spoken to by AI, not just by other software. Practically, the question to put to any vendor pitching you is no longer "do you have an API". It's "can something else drive you". Those are different answers, and within a couple of years the second one decides which of your systems can join a workflow and which can't.
Heine told the panel Netwealth now thinks of itself "far more as a tech business than a wealth business", and described an obligation to build "that data spine that actually allows you to then have on-ramps and off-ramps for different solutions" so as "to avoid the data having to be re-keyed throughout that whole ecosystem". He floated starting right back at the fact-find, so nothing gets re-entered at application. Pettiona was blunter about the motive: "The platforms are rapidly commoditising."
Re-keying is the single most reliable tell I use when I map a practice. Wherever a human types the same data into a second system, there's automatable work sitting there, every time. So it's worth noticing that the platforms have landed on the same diagnosis and are offering themselves as the cure. For the boring plumbing, that's good news. It also raises a question worth asking before you lean in: if your data spine belongs to your platform, how easily do you leave? A commoditising business goes hunting for stickiness, and owning the connective layer is the stickiest thing on the table. Notice that Padua, one item up, is making the same play from the other direction. Two of them building the same road is better for you than one. Use it, by all means. Just go in knowing what it would cost to unpick.
That's the fortnight. Three items that all turned out to be about the same thing, which I hadn't planned. If you reckon I've read any of it wrong, hit reply and tell me. Past issues live here.
From Bax Advisory
Three minutes, twelve questions, and a scored report telling you where your practice actually stands. Free.
Take the checkSixty minutes. You leave with an AI policy, a tools register and a clear RACI, yours to take to your licensee.
Book the sessionThirty minutes, no pitch deck, no obligation. A straight read on where AI fits in your practice.
Book a discovery callGet It In Your Inbox
One email a fortnight. Three or four things worth knowing, read in five minutes. Unsubscribe whenever you like.
We'll email you a confirmation link first, so nobody ends up on the list by accident. Your address is used for the Brief and nothing else.