The short answer
Paraplanners can use AI for the preparation behind review packs and client documents: reading statements, filing them to the right accounts, calculating returns and writing a first draft in the firm’s template. Advice, suitability and sign-off stay with people. Start with one task, give the AI your template and past signed-off examples, agree data handling with compliance, test it on past cases, and keep a person approving every output.
- Reads valuations and statements
- Files each document to the right account
- Identifies holdings by ISIN
- Calculates returns from the data
- Drafts the pack in your template
- Lists anything it could not verify
- Advice and recommendations
- Suitability
- Checking flagged figures
- Edits to the draft
- Final approval
What AI is good at in paraplanning
Much of a paraplanner’s week goes on preparation that has to happen before anyone can review anything. That preparation is where AI is useful today:
- Reading documents. Platform valuations, transaction histories, fact finds and meeting notes, in whatever format each provider sends them.
- Filing and matching. Working out which account and which account holder each document belongs to, and identifying every holding by ISIN, or provider fund code where there is none.
- Calculating from data. Working out returns from the client’s valuations and cash flows, rather than copying a figure off a statement.
- Writing a first draft. Commentary, summaries and letters in your template and your firm’s voice.
- Spotting gaps. Listing anything it could not find or verify, so the reviewer knows exactly where to look.
What should stay with people
Advice, suitability, recommendations and final approval. AI drafts and prepares; a paraplanner or adviser decides what reaches the client. The same goes for judgement about a client’s circumstances, and for anything the AI has flagged as unverified.
Good first tasks
| Task | Why it is a good place to start |
|---|---|
| Review pack preparation | Repetitive, document-heavy, and always checked by a person before it goes out |
| Meeting note summaries | Turns a transcript or notes into actions and a file note in your format |
| Fact find data entry | Moves information from documents into your system without re-keying |
| Fund research summaries | Pulls fund data together by ISIN for a first read |
| Client letter drafts | A first draft in your template, edited and approved by the adviser |
Related guides
Step by step
- 01
Time your current process
Take five recent reviews, simple and complex, and note how long each stage took: gathering, reconciling, re-keying, calculating, writing and formatting. This shows where the time goes, and gives you a baseline to measure against.
- 02
Choose one task to start with
Pick a task that is repetitive, document-heavy and checked by a person anyway. Review pack preparation is usually the strongest candidate. Avoid starting with anything that involves a recommendation.
- 03
Gather your firm’s inputs
Collect your review template, house view, preferred tone and a handful of past packs you have already signed off. The AI should produce your document, in your format, not a generic one.
- 04
Agree data handling with compliance
Decide where client data will be held (UK or EU hosting, or inside your own systems), put a data processing agreement in place with every AI supplier, and confirm client data is not used to train AI models. Never paste client data into a free consumer chat tool.
- 05
Make the AI show its working
Every figure should link back to the document it came from. Returns should be calculated from valuations and cash flows rather than copied off statements, with any provider-reported figure labelled as such. Ask for a list of anything it could not find or verify.
- 06
Test on past cases before going live
Rebuild a set of reviews you have already signed off, including the awkward ones, and compare the results. Agree the bar in advance: for example no material omissions, presentation your team accepts, and a time target you set.
- 07
Keep a person in charge
A paraplanner checks and edits every output, and a named person, usually the adviser, approves it before it reaches a client. Log every change so you can see what was altered, by whom and when.
- 08
Roll out gradually
Start with one adviser’s clients, then widen. Revisit the setup whenever your template, providers or house view change, and spot-check finished packs regularly.
Common mistakes
- Offshore bond growthProvider’s own figure, not recalculatedFrom statement
- Joint investment account, Whitlock familyHolder not confirmed in the documentsUnverified
- Return after June top-upCalculated from valuations and cash flowsCalculated
- Pack approvalNo approver assigned yetNeeds a person
- Copying growth figures straight off provider statements instead of calculating them from the client’s own valuations and cash flows.
- Mixing up linked family accounts, joint holders or trusts when documents are filed.
- Offshore bonds and multi-provider portfolios that report a single growth figure and hide the detail underneath.
- Pasting client data into a free chat tool with no agreement on where it is stored or how it is used.
- Choosing a tool that handles simple no-change reviews well when most of your reviews are not simple.
- Letting drafts reach clients without a named person approving them.
Questions
Can AI do paraplanning?
AI can do much of the preparation: reading valuations and transaction histories, filing them to the right accounts, identifying holdings, calculating returns and drafting documents in your template. The judgement, the checking and responsibility for the advice stay with the paraplanner and adviser.
Will AI replace paraplanners?
It takes over repetitive preparation such as gathering, re-keying and formatting. Checking figures, understanding the client and making sure a document says the right thing still need a paraplanner. The aim is to give paraplanners more time for that work.
Is it safe to use ChatGPT for client reviews?
Not by pasting client data into a consumer account. Use a setup where your compliance team has agreed where data is held, a data processing agreement is in place, and client data is not used to train models. Check your firm’s own policies first.
What should a paraplanner check in an AI-drafted review pack?
Check figures against their linked sources, look at everything the AI flagged as unverified, confirm documents were filed to the right accounts and account holders, and make sure the commentary fits the house view and the client’s circumstances.
Where should a paraplanner start with AI?
Start with one repetitive, document-heavy task that a person checks anyway, such as review pack preparation. Time how long it takes today, test the AI on past cases you have signed off, and only go live when it meets the bar you agreed.