A financial planning practice with a modest client base produces a surprising number of documents in any given month. Not one type: six or seven. Each serves a different purpose, carries a different compliance weight, and takes a different amount of time to produce. Some are well suited to AI-assisted drafting. Others require more of the adviser. Knowing which is which is the starting point for understanding where a document pipeline actually helps your practice. This piece maps each of them, and is a companion to the fuller guide on how financial planners are automating compliance documents with AI, which covers the workflow, the compliance framework, and the time and cost case end to end.
The document load in a typical month
These are the main document types a financial planning practice produces regularly, with honest time estimates based on paraplanning industry data.
Statement of Advice (SOA). The most time-intensive document a financial planning practice produces. An SOA is required for personal financial advice to retail clients, per ASIC’s regulatory framework. According to paraplanner.com.au, a straightforward risk SOA takes around 2.5 hours to produce; a complex multi-entity SOA can take nine hours. At outsourced paraplanning rates of $100 to $150 per hour, each SOA costs $250 to $1,350 to produce. For a practice writing ten to twenty per month, this is the single largest documentation cost in the business.
Record of Advice (ROA). Used for follow-up advice where a previous SOA has already been provided. Shorter and less formal than an SOA, but still required documentation. An ROA for a routine portfolio review or product switch is typically one to two hours. Practices that run regular review programs produce a lot of these.
Financial Services Guide (FSG). A one-off document that explains your services, fees, and how you handle complaints. Requires updating when services or fees change. Lower time cost than an SOA; higher stakes if it is out of date.
File notes. A record of every client meeting or significant conversation. Required as part of your obligation to document the basis for advice. Time cost per file note is lower than an SOA but the volume is high: every client interaction generates one. Practices often underestimate how much cumulative time this takes.
Annual review letters and client correspondence. Review invitations, portfolio summaries, product change notifications, general client communication. Individually low time cost; collectively significant across a full client base.
The common feature across all of these: they follow established structures. The content changes with each client. The form largely does not.
Which documents AI handles well
AI-assisted drafting performs best on documents that follow consistent patterns and where the cost of a minor error in the draft is low because a human is reviewing before anything is sent.
File notes are the strongest current use case. Meeting recording tools transcribe the conversation; AI structures the output into a file note in the practice’s own format. What used to take twenty minutes of writing after a meeting takes two or three minutes of review. Across a full week of client meetings, this is a meaningful time recovery.
ROAs for routine follow-up advice are well suited to AI drafting when the underlying situation is familiar: a portfolio rebalance for an existing client, an annual review recommendation, a product switch within an established strategy. The AI drafts from the structured inputs; the adviser reviews and signs off. For practices running regular review programs, this is where volume savings are largest.
Client correspondence and review invitation letters have high template consistency and low compliance risk if the draft has a minor wording issue. AI handles these cleanly.
Template-driven sections of SOAs, such as scope of advice statements, service fee disclosures, and disclaimer blocks, can be generated reliably and reviewed quickly. The variable sections (the actual advice, the product recommendations, the client-specific reasoning) still need the adviser.
Which documents benefit from more human input
The pattern is consistent: AI drafts well when the document follows established structure and the inputs are clear. It produces drafts that need more work when the situation is genuinely novel or the judgement call is complex.
Complex multi-entity SOAs involving corporate trustees, SMSFs, blended family structures, or overlapping advice areas produce more variable AI output. The structural sections draft cleanly. The advice reasoning sections, which need to accurately reflect a complex client situation and hold up to scrutiny, require the adviser’s full attention. AI produces a starting point; the adviser does substantive work on it.
SMSF advice documentation involves specific legislative and compliance considerations that fall outside the patterns most AI tools are trained on. Draft quality varies more; review time increases accordingly.
Any advice involving novel client circumstances, specifically a situation the practice has not encountered in substantially the same form before, is where AI output is least reliable as a starting point. The tool is working from patterns. Genuinely new situations break those patterns.
The honest framing for all of these: AI can be trained to flag when a document departs from its training data, and can incorporate client-specific data and practice input to improve draft quality over time. But human review before a document leaves the practice is the standard, not the exception. That is not a limitation specific to AI. It is the appropriate standard for any document that carries compliance weight and goes to a retail client.
What changes in your week
The time recovery from AI-assisted document production is not evenly distributed. It shows up most clearly in the documents that have the highest volume and the most consistent structure: file notes, ROAs, client correspondence, and the boilerplate sections of SOAs.
For a practice writing fifteen SOAs per month alongside twenty file notes and ten review letters, reducing average production time by ninety minutes across the full document mix returns roughly fifty hours per month. That is time that can go to client-facing work, complex cases, or simply finishing on time rather than after hours.
What does not change: the adviser’s responsibility for every document that leaves the practice. Review is built into the workflow, not optional. The value of AI-assisted drafting is that the review takes less time because the draft is already structured correctly, not that the review can be skipped.
Who this does not suit
Very low document volume. If a practice produces fewer than five to eight documents per month across all types, the setup and configuration effort does not pay back quickly. The tool is designed for recurring volume, not occasional use.
No consistent templates. A document pipeline trained on practice documents needs those documents to be reasonably consistent. A practice where three advisers write SOAs in three different styles will get variable output until the templates are standardised. Getting that right first is worth doing regardless. It is useful practice hygiene. But it is a prerequisite.
Practices not yet set up for digital review workflows. If the current process involves printing documents and reviewing them on paper, adding an AI drafting step to the front of that workflow does not change much. The time saving is realised when review and sign-off happen efficiently too.
The map matters more than the tool
Knowing which of your documents AI handles well and which ones still need you is more valuable than any individual tool recommendation. The practices that get the most from AI-assisted drafting are the ones that have mapped their document load clearly: what they produce, how often, how long each type takes, and where the time is actually going.
Once that map exists, the decision about where to deploy a document pipeline is straightforward. The documents that follow consistent patterns and appear frequently are the starting point. Everything else follows from that.
See how Imprint builds a document pipeline for your practice: upgraded.au/ai-tools/imprint