Use cases/Financial services/Advice file drafting
Advice file drafting for Australian advice practices
Advice file drafting means a model builds the first draft of a Statement of Advice from the fact find, the platform data and your approved product list, in the licensee template. The adviser reviews the draft and owns the advice. What goes is the rebuilding of the same facts into a new document for every client.
Talk to our team →- Supports
- Paraplanner
- Reads
- Fact find, Platform data, Approved product list, Client agreements, Licensee checklist
- Runs on
- Your servers, on premises
- Connects to
- Iress Xplan, AdviserLogic, Midwinter, HUB24
- Decides
- A person, every time
Architecture
Everything runs on your servers
We build it on an open-weight model such as Llama or Mistral, served with vLLM on NVIDIA GPUs in your server room, data centre or colocation. Retrieval runs on PostgreSQL with pgvector, and the system reads from your advice software and platforms. No client data goes to a public AI service.
- SourcesFact find / Platform data / Approved product list / Client agreements / Licensee checklist
- Your serversRead the fact find → Check product list → Draft the advice
- ReviewParaplanner
- OutputStatement of Advice drafts / Fee consent tracking / Compliance review of each file / Adviser review and ownership
It takes four jobs off the paraplanner
- 01
Statement of Advice drafts
The draft is built from the fact find, the platform data and the approved product list in the licensee template. Each figure shows the document it came from.
- 02
Fee consent tracking
Consent forms are drafted from the fees and services in each client agreement. Every renewal date is tracked and unsigned consents are listed before the window closes.
- 03
Compliance review of each file
Each advice file is read against the licensee checklist. Missing documents and gaps between the fact find and the advice are listed for the reviewer.
- 04
Adviser review and ownership
The adviser reads the draft, changes it and signs it. Nothing goes to a client without that sign off.
It goes live in five steps
- Step 01
Load your templates and rules
The licensee template, the approved product list, the file review checklist and a set of past advice files, structured and attached to the client they belong to.
- Step 02
Test against past advice files
Run the draft over files already approved and compare. You see where it differs from the adviser's version and on what type of advice, and that sets the review rules.
- Step 03
Draft from the source records
The Statement of Advice is built from the fact find, platform data and approved product list, with the source beside each figure.
- Step 04
Adviser reviews and owns the advice
The adviser reads every draft, changes what needs changing and signs it. The best interests judgement stays with them.
- Step 05
Check the file and track consents
Each file is read against the licensee checklist before it closes. Fee consent dates are tracked and unsigned items are listed.
Software or a custom build
If your practice runs on Xplan or AdviserLogic, check the document tools inside them first, and ask the vendor what AI features are on the roadmap. A custom build earns its cost when your licensee template is specific to you, when advice data sits across several platforms, or when client data cannot go to a vendor cloud. We will say in discovery if the tool you already pay for is enough.
Iress Xplan / AdviserLogic / Midwinter / HUB24 / Netwealth / BT Panorama / Class
In practice
The system drafts and checks from your own records. An adviser or a compliance officer approves each output.
The record is the design constraint. ASIC sets out what a Statement of Advice must contain. Records of personal advice must be kept for at least seven years. The ongoing fee consent rules set a renewal window and a record keeping period, which is why we track each date. For how a wealth manager gets a technical opinion on AI tools before signing, see an example engagement.
The method is in the library. Read where to put the human for the review step and evaluating and observing LLMs on your own servers for how we test a draft against past files. Our LLM integration and RAG and private AI services cover the build, and the wider picture is on the wealth management page.
Questions
Can AI write a Statement of Advice?
It can draft one. The draft comes from the fact find, the platform data and your approved product list, and each figure shows its source. The adviser still owns the advice and the best interests duty, and nothing goes to a client without their sign off.
How does fee consent tracking work?
The system reads the fees and services in each client agreement and tracks every renewal date. It drafts the consent form and lists unsigned consents before the window closes. A person sends each form. If no valid consent is signed by the end of the window, the arrangement ends by law.
Where does client data go?
It stays in your environment. The model runs on your own servers, and no client data is sent to a public AI service. Staff see only the client files they can already open in your advice software.
Do we need to change our advice software or platform?
No. What we build reads from Xplan, AdviserLogic and the platforms you already run, and puts drafts back where your team works.
Industries where this comes up
More financial services use cases
Document search / Credit file assembly / AML alert triage / Claims triage / Month-end close / Audit evidence matching / Lease abstraction / KYC and onboarding / Complaints handling / Regulatory reporting / Underwriting submissions / Board and management reporting
Before you call
Tell us where your hours go.
We map the work, size the hardware and price what is worth automating.
Talk to our team →