A Statement of Advice takes a qualified paraplanner 2.5 to 9 hours to produce at $100 to $150 an hour outsourced. Most financial planning practices produce ten to twenty per month. AI document pipelines are cutting that production time substantially. Here is what that actually looks like in a financial planning workflow, and what ASIC’s framework requires when you use these tools.
The document production problem in financial planning
A financial planning practice does not produce one type of document. It produces several, continuously.
Statements of Advice are the most time-intensive: according to paraplanner.com.au, a straightforward risk SOA takes around 2.5 hours; a complex multi-entity SOA can take nine hours. At outsourced paraplanning rates of $100 to $150 per hour (as of May 2026), that is $250 to $1,350 per document. A practice producing fifteen SOAs per month is looking at $3,750 to $20,250 per month in paraplanning costs at the high end, or $45,000 to $243,000 annually.
For a document-by-document breakdown of which types AI handles well and which still need the adviser, this companion piece maps the full document load for a typical financial planning practice.
Records of Advice sit below SOAs on the complexity scale. They are used for routine follow-up advice where a previous SOA already exists, and they are shorter. But they are still required documentation. Financial Services Guides, file notes from every client meeting, annual review letters, and client correspondence all stack up alongside them. These are not optional documents; they are the operational and compliance backbone of any practice.
The problem is not that advisers cannot produce these documents. It is that the production time comes directly out of time that could be spent with clients or on the judgement-heavy work that actually requires a qualified adviser.
What AI tools are actually doing in this context
AI document tools in the financial advice space are not making advice decisions. They are handling the structural drafting work that follows a decision the adviser has already made.
The workflow looks like this: an adviser meets with a client, records notes or uses a meeting-capture tool, and feeds structured information into the AI pipeline. The AI drafts the document (SOA, ROA, file note, review letter) in the practice’s own format, using the practice’s own language and templates. The adviser reviews, edits where needed, and signs off.
Several tools operate in the Australian market at various stages of maturity. Claras focuses on file note generation from meeting recordings. Paradino targets SOA drafting specifically. Imprint, Upgraded’s document pipeline tool, is trained on a practice’s existing documents rather than a generic template, which means the output sounds like the practice and not like a software product.
The common thread: AI handles the structural drafting. The adviser handles the judgement, the review, and the sign-off. These are not interchangeable roles.
What ASIC’s framework requires when you use AI
Australian financial planning practices that hold an Australian Financial Services Licence (AFSL) operate within a regulatory framework that does not change when AI tools enter the workflow.
In October 2024, ASIC published REP 798, a review of how AI is being adopted across financial services and credit licensees. The finding that matters most for financial planners: ASIC concluded that its existing regulatory framework is technology-neutral. The obligations that govern how you provide financial services apply equally whether you use AI or not.
That means three things in practice.
First, the obligation to provide services “efficiently, honestly and fairly” (Corporations Act 2001) applies to AI-assisted document production. If an AI tool generates advice that is not suitable for the client’s circumstances, the adviser remains responsible. ASIC has specifically warned that AI models can produce outcomes that are difficult to explain and can treat consumers unfairly if the underlying model is not properly governed.
Second, human oversight is not optional. REP 798 makes clear that meaningful human control over AI output is a compliance expectation, not a courtesy. A licensed adviser must review the document before it is provided to a client. That review needs to be substantive: confirming that the advice is suitable, accurate, and reflects the client’s actual circumstances. Not a cursory sign-off.
Third, record-keeping obligations extend to AI use. Under section 286 of the Corporations Act, financial records must be retained for a minimum of seven years after the transactions they cover. ASIC’s expectation, as described in REP 798, is that licensees maintain records of their AI governance arrangements, including documentation of how AI tools are used, monitored, and reviewed. Keep records of how documents were generated, not just the documents themselves.
None of this prevents AI from being useful in a financial planning practice. It defines where human judgement stays in the loop and what documentation you need to show it was there.
The time and cost case
The business case for AI-assisted document production is clearest when you run the maths on your own practice’s output volume.
Take a practice producing fifteen SOAs per month, using an outsourced paraplanner at an average of $125 per hour, with an average production time of four hours per document. That is $7,500 per month in paraplanning costs for SOAs alone, or $90,000 per year.
AI-assisted drafting does not eliminate that cost entirely, because the adviser still reviews each document. But it changes what the review involves. Instead of a paraplanner writing from scratch and an adviser reviewing a finished document, the AI produces a structured draft in minutes and the adviser reviews and edits. Industry experience from document pipeline providers suggests meaningful time reductions, but conservative figures are more useful than vendor claims here. Even a reduction of one to two hours per document at scale adds up to significant operational savings.
At fifteen SOAs per month, saving two hours per document frees thirty hours of paraplanner time. At $125 per hour, that is $3,750 per month back into the practice, enough to cover an AI document tool and leave a margin.
The case is stronger if your practice is growing. Adding clients without adding paraplanning costs in proportion is the operational lever that AI-assisted document production actually provides.
What changes in your week, and what does not
The honest version of this is that AI-assisted document production changes where your time goes, not how much compliance your practice carries.
What changes: the time from client meeting to draft document. File notes can be generated from meeting recordings rather than written afterwards. SOA templates are populated from structured client data rather than assembled from scratch. The blank-page phase of document production largely disappears.
What does not change: the adviser’s responsibility for every document that leaves the practice. Review is not a formality. ASIC’s REP 798 is explicit that governance must keep pace with AI adoption, and that a human reviewer must understand enough about the tool’s behaviour to catch errors and anomalies, not just read the output and approve it.
This matters practically: you need to know what your AI tool does when client circumstances fall outside its training data. Most tools handle routine cases well. Novel situations (a client structure the tool has not seen before, advice that requires unusual judgement) are where human review earns its place.
The workflow that works is AI for the structural draft, adviser for the judgement layer, and clear documentation that both happened.
Who this does not suit
Worth being direct about where AI document tools are not the right fit.
Practices with fewer than five to eight SOAs per month. At low volume, the return on setting up and maintaining an AI document pipeline is modest. The operational savings are real but not transformative, and the configuration investment may not be worth it until volume grows.
Highly bespoke advice structures. AI tools work well when documents follow recognisable patterns. A practice specialising in unusually complex structures (SMSF strategies with multiple related entities, intricate business succession arrangements, or advice involving overlapping legislative frameworks) will find that AI drafting requires more rework than it saves. The tool is trained on patterns; genuinely novel situations break those patterns.
Practices not yet on consistent templates. A document pipeline trained on a practice’s existing documents needs those documents to be reasonably consistent in structure and language. If the practice currently produces SOAs in five different formats depending on who wrote them, the AI output will reflect that inconsistency. Getting templates standardised first is not a barrier. It is useful work regardless. But it is a prerequisite.
Advisers who have not read REP 798. Not a technology problem. If a practice adopts AI document tools without updating its governance, risk, and compliance frameworks, ASIC has already flagged that as an area of concern. The tools work; the governance arrangements need to keep up.
The compliance framework is workable
ASIC’s position on AI in financial advice is not a prohibition. It is a clarification: the existing obligations apply, the adviser remains accountable, and governance must keep pace with the technology being used.
For a practice doing ten to twenty compliance documents per month, the time and cost case for AI-assisted drafting is hard to ignore once you run it against your own numbers. The question is whether your document pipeline is trained on your practice’s language and structure, or whether you are working with a generic template that sounds like it came from a software product.
See how Imprint generates compliance documents trained on your practice’s templates: upgraded.au/ai-tools/imprint