Give every fact one entry point, usually a form the client completes, and let every other system read from it. The form feeds the CRM through Zapier or Make, the planning software takes the CRM record where the two connect, an AI step drafts the file note, and consent boxes decide whether it runs.
Expect re-keying to survive in two places: platform application forms, and any planning tool that offers look-up rather than client creation. Wealthbox, for example, describes its eMoney and MoneyGuide connections as account look-up and single sign-on, and its RightCapital link as syncing notes and tasks. Mapping every step first shows exactly where the gaps are and what to do about each.
Count how many times each client is keyed today
Before you automate anything, time one real onboarding and write down every place a human typed the client's details. The example here is a three-adviser firm onboarding about 20 new clients a month. Its audit looks like this, and yours will differ.
| Where the details are typed | Minutes per client | Typical fault |
|---|---|---|
| Enquiry log (spreadsheet) | 5 | Phone number with a digit missing |
| CRM contact and household | 10 | Partner's date of birth typed on the client's record |
| Planning software client record | 20 | Income entered gross in one place, net in another |
| Platform or product application | 15 | Address copied from an old record |
| File note for compliance | 20 | Written from memory days later |
| Total | 70 |
Seventy minutes of typing per client, five separate times, is the cost of re-keying, and the last column is the real cost: five chances to introduce a different version of the truth. The aim is to cut the typing to one pass done by the client, and to turn the adviser's time into checking.
The entry-point map: one place to type each fact
Draw the onboarding steps down the left side and, for each, name the single place where its data is typed. Every other system either receives a copy automatically or looks the data up. If a step has two places where a human types, you have found a re-keying point.
| Step | Typed once, in | Everything else reads it from | How it moves |
|---|---|---|---|
| 1. Enquiry and intro call | Enquiry form or booking page | CRM contact in a "lead" stage | Zapier finds or creates the contact and adds a call-prep task |
| 2. Fact-find | Client fact-find form, completed by the client | CRM fields, household, a PDF copy, the AI note | Zapier updates the contact, builds the household and adds the answers as a note |
| 3. Identity and source-of-funds checks | The identity-check provider's own upload page | CRM status field and a link, never the document | Provider status triggers a contact update |
| 4. Agreements and consents | One e-signature envelope | CRM stage, signed PDF link, compliance file | Completed envelope triggers a note and a stage change |
| 5. Platform or product application | The platform's own form, pre-filled from the CRM where it can be | Nothing; this is the end of the chain | Pre-fill tool, or a one-page application data sheet from the CRM |
| 6. Planning software client record | The CRM record (already typed in step 2) | Planning software | Native CRM link or import file, tested with a dummy client |
| 7. File note | AI draft from the CRM record, edited by the adviser | CRM note and compliance file | An AI step creates a draft marked for review |
| 8. Welcome pack and first meeting | A CRM stage change | Client's inbox | A Wealthbox workflow or a Zap sends the email and booking link |
Three rules keep the map honest. A field has one owner: once the form captures date of birth, nobody types it into the CRM. Documents stay where they are uploaded, and only a status and a link travel onward. And every row names a mechanism, so "Zapier updates the contact" is a real step you can test, while "it syncs" is not. Where several onboarding tools might feed the same CRM, what is built in and what is bolted on in an adviser CRM helps you choose between native features and add-ons.
The Complete Excel Course with AI Integration
From formulas and pivot tables to Power Query, DAX, VBA and AI with Copilot, ChatGPT and Claude — hands-on from the first lesson.
By Sagnik Bhattacharya and Virginia Thorn, Primetime Emmy Award winner
Buy nowDesign the fact-find form so mistakes cannot be typed
The cheapest error check is a form that refuses bad input. Free text is the enemy of automation: a planning import cannot read "about 60k-ish", and neither can the AI step. Build every field to produce something a machine can map.
| Field | Input type | Rule | What it prevents |
|---|---|---|---|
| Date of birth | Date picker | Required, in the past, adult age check | Day and month swapped, impossible dates |
| Email and phone | Email and phone fields | Required for each adult, own address each | Letters in numbers, partners merged into one contact |
| Employment status | Dropdown | Employed, self-employed, retired, not working | Free text nobody can map |
| Annual income | Number | Label says "before tax", fixed currency | Gross entered as net |
| Assets and liabilities | Repeating rows | One row each: type, provider, balance | A paragraph the import cannot split |
| Goals | Short text plus target year | A year is required for each goal | The AI inventing a timeline |
| Risk questionnaire | Scored multiple choice | The form stores the score and the label | Anyone retyping or reinterpreting it |
| Documents | No upload field | A link to the identity provider instead | Copies of ID documents in the form tool |
The email row hides a trap worth testing. Couples often share one address, and a "find or create" step that matches on email can merge two partners into a single CRM contact. Submit a dummy couple who share an email and see what happens. If the CRM merges them, give each adult their own email field, match on something else too, or create the partner explicitly.
From CRM to planning software: what moves and what still gets typed
This hop is the weakest link, so check it with each vendor before promising anything. Wealthbox's integrations page says its eMoney and MoneyGuide connections give account look-up and single sign-on, which is not the same as creating a client record. It says its RightCapital link syncs notes and tasks. RightCapital's own integrations page lists Redtail, SmartOffice and Wealthbox among the CRMs it works with to import client data, but it does not say which fields move. Wealthbox also lists Quik! and Docupace for pre-filling forms from contact information.
The practical test takes an hour. Create a dummy household in the CRM with a distinctive value in every field (date of birth 11 November 1911, income $123,456) and push it through whatever link exists. Then list each field that arrived, each that did not and each that landed in the wrong place. The fields that did not arrive are your remaining re-keying, and you can time them. If your planning software only supports look-up, ask the vendor whether an import file can create clients, and put that file's template in the map. For help with the vendor conversation, see what to test first in AI financial planning software.
Form to CRM: the Zap, step by step, with task counts
The example uses Wealthbox because Zapier's own page for it lists the steps clearly. If you use another CRM, open its Zapier or Make directory page and look for the equivalents. Wealthbox's actions include Find or Create a Contact, Update Contact, Create Household, Add Member to Household, Create Note, Create Task and Start Workflow. Wealthbox's pricing page lists workflows and automations from its Pro plan, at $75 per user per month (it does not state the billing term). Here is the Zap for step 2, the fact-find, in order:
- Trigger: a new submission on the fact-find form. Triggers do not count as tasks.
- Find or Create a Contact, using the email address, so a returning enquirer never becomes a duplicate (1 task).
- Update Contact with the fact-find fields (1).
- Create Household, then Add Member to Household for a partner, on joint clients (2).
- Create Note holding the full set of answers (1).
- Start Workflow, the "new client onboarding" checklist (1).
- Filter: continue only if the AI-consent field is yes. Filters do not count as tasks.
- AI step: draft the file note (1, 3 or 5 tasks depending on the model tier; Zapier's help centre says new steps on paid plans start on the 5-task Premium tier, so count 5 unless you change it).
- Create Note with the draft, marked for review, and Create Task for the adviser (2).
That is 13 tasks a client. Add a small Zap for the e-signature completion (3 tasks) and one for the identity-check status (1), and 20 new clients a month use about 340 tasks. Zapier's Professional plan covers that in its 750-task tier at $19.99 a month on annual billing, or $29.99 if paid monthly. The Free plan cannot run it, since it allows 100 tasks a month, two-step Zaps only and no filters. Because the consent filter sits after the Start Workflow step, a client who declines the AI step still gets the rest of onboarding, and the workflow's own "write the file note" task covers the manual route.
Make is the alternative. Its Free plan gives up to 1,000 credits a month, an AI step can use more than one credit, and Make's pricing page showed no paid prices when I checked, so read it directly before you plan. Wealthbox's integrations page still lists Make under its earlier name, Integromat. For the wider pattern, five onboarding automations for advisers on Zapier shows others to add later.
Consent capture: separate boxes that can switch the AI off
Consent belongs in the same form as the facts, so it is captured once with a timestamp. Ask your compliance adviser for the exact wording your regulator and data-protection rules require. This skeleton shows the structure, not the legal text.
CONSENTS (every box unticked by default, each separate)
[ ] I have read the privacy notice at [link] and understand how
[firm] uses my information to provide advice.
[ ] I agree that [firm] may send me documents electronically.
[ ] I understand that [firm] uses an AI tool to draft a summary of
the information I give, that an adviser reviews every summary,
and that I can ask for a person to write it instead.
[ ] (Optional) I would like to receive newsletters and invitations.Four design points matter. Each box is its own consent, with the marketing one optional and never bundled. Each answer is copied into a CRM field (privacy notice acknowledged with a date, e-delivery yes or no, AI-assisted note yes or no, marketing yes or no), so the Zap's filter reads a field rather than parsing text. The submission record and the e-signature audit trail are your evidence, so keep both. And the AI line must describe what really happens: an adviser reviews every summary. If a client cannot use the form, the adviser completes the same form with them on screen. That keeps one entry point, with no side spreadsheet.
The AI file note: draft, flag gaps, never recommend
The AI step has one job: turn the client's own answers into a draft the adviser can check in minutes. It must not advise, interpret or fill gaps. Here is a prompt that enforces that.
Write a DRAFT file note for an adviser to review, using ONLY the
data below.
Sections: Client and household / Goals and timeline / Income,
assets and liabilities (as stated) / Attitude-to-risk questionnaire
(score and label exactly as given) / Gaps and inconsistencies /
Documents still needed.
Rules: no recommendations, no suitability opinion, no calculations
beyond adding figures given. Do not infer age, retirement date, risk
tolerance or intentions. Write [CHECK: ...] for any blank or
inconsistent field. Begin with "DRAFT, NOT REVIEWED".A good run, with illustrative figures, reads like this.
DRAFT, NOT REVIEWED. Client and household: joint clients, two dependants (ages given as 9 and 12). Goals and timeline: retire in about 12 years, help with university costs in 6 years, [CHECK: no target retirement income given]. Income, assets and liabilities (as stated): combined salary $138,000, savings $42,000, workplace retirement accounts $310,000, mortgage $215,000, [CHECK: partner's employment status blank]. Attitude to risk: score 52, label "moderate". Gaps and inconsistencies: partner's status is blank while the salary field is combined. Documents still needed: proof of address; recent retirement account statements.
Compare that with a first-run draft and what the adviser corrected. Each fix became a line in the prompt.
| First-run wording | What the data said | Fix in the prompt |
|---|---|---|
| "Client plans to retire at 65" | "In about 12 years"; no age was supplied, so the tool guessed one | "Do not infer age or retirement date" |
| "Risk profile: balanced" | Score 52, label "moderate" | "Score and label exactly as given" |
| "Healthy savings buffer" | Only a balance, with no target to compare against | "No evaluative language" |
| "Client income $138,000" | The form field was combined household income | "As stated", plus a clearer field label on the form |
Keep data to the minimum the note needs. Map only the fields in the sections above, and leave out identity numbers, account numbers, passwords and any uploaded document. Zapier's AI step counts 1x tasks with your own API key, so that route lets you choose a provider whose data terms you have read. Whichever route you take, read where the data goes before sending the first real client. The adviser then edits the draft, removes the DRAFT line and records that the note was reviewed. For the compliance side of AI-written advice documents, can AI draft a suitability report that passes compliance review covers the line between a summary and advice.
Error checks: ten places where onboarding data goes wrong
| Check | Where it sits | What it catches |
|---|---|---|
| Required fields and format rules | The form | Blank dates of birth, emails missing an @, phone numbers with letters |
| Date pickers, not typed dates | The form | Day and month swapped (05/06 read two ways) |
| Find or Create a Contact | First Zap step | Duplicates when a client enquires twice |
| Unique-value mapping test | Before go-live | A partner's date of birth landing on the client's field |
| Consent filter | Before the AI step | The AI running on a client who declined |
| Cross-field sense flags | The AI prompt | "Retired" with a salary, assets smaller than the listed accounts |
| Failure alerts to a monitored inbox | Zap settings | A run that errors halfway and leaves a half-built household |
| Weekly three-way count | A spreadsheet | Submissions, CRM contacts and planning records that do not match |
| Adviser review sign-off | CRM task | Any AI wording that slipped past the prompt |
| Plan limit alerts | Zapier account | A month-end surge hitting the task cap (Zapier holds the runs rather than losing them, but onboarding stalls) |
Two of these deserve a second look. Zapier turns a Zap off automatically when it errors on 95 per cent of its runs and has run more than 20 times in the past seven days (Team and Enterprise accounts get a short grace period first), so a dead automation can look exactly like a quiet week, which is why the weekly count exists. And the unique-value test beats any other mapping check: submit a dummy form in which every field holds a different, recognisable value, then see where each value lands. Review steps that pause a run for a person are also worth knowing about: Zapier's Human in the Loop offers "End run" or "Skip and continue" as timeout choices, while Make's equivalent app is Enterprise-only and in closed beta.
A four-week build and what it saves a three-adviser firm
- Week 1 (about 6 hours): audit and map. Time one real onboarding, draw the entry-point map, name the owner of each field and agree the consent wording with compliance.
- Week 2 (about 10 hours): form and core Zap. Build the fact-find form, connect it to the CRM and run the unique-value test until every field lands correctly.
- Week 3 (about 8 hours): the rest of the chain. Add the e-signature and identity-status Zaps, test the planning software hop with a dummy household and write the AI prompt and the adviser's review step.
- Week 4 (about 6 hours): run in parallel. Onboard three real clients through both the old and new routes, compare field by field, fix the differences and then switch. Start the weekly three-way count.
That is about 30 hours of build. After it, the typing per client falls from 70 minutes to about 25, and what remains is checking. These figures are illustrative.
| Remaining firm time per client | Minutes |
|---|---|
| Review the submitted fact-find | 5 |
| Planning software hop (import, link test or typing the remainder) | 5 |
| Platform application data sheet | 5 |
| Review and edit the AI file note | 8 |
| Spot checks on mapping and consents | 2 |
| Total | 25 |
The saving is 45 minutes a client, so 20 clients a month frees 900 minutes, or 15 hours. At an internal cost of $60 an hour (use your own), that is about $900 a month. Against it, Zapier Professional costs $19.99 a month on annual billing, and the form tool costs $0 on a free tier or $39 on Jotform's Bronze plan, so running costs stay under about $60 a month. The 30-hour build is $1,800 at the same rate, which the saving repays in roughly two months. Be honest about one adjustment: if 6 of the 20 clients need an adviser to complete the form with them for 15 minutes each, that adds 90 minutes, and the saving falls to 13.5 hours.
The numbers matter less than the habit they enforce. Each time someone proposes a new tool or a new field, ask the same question: where is this typed, and by whom? If the answer is "in two places", fix that before anything else is automated. If you would like to walk your own onboarding steps through that map with someone, that is what my AI implementation consultation covers, and you can start from the contact page. When the agreement is the next step to tackle, drafting the client agreement with AI and then checking it is a good place to go, and the identity step is covered in KYC checks that stay human in a small advice firm.
Questions advisers ask about automating onboarding
Can the AI step make suitability or risk judgements?
No. Keep it to summarising what the client wrote, flagging gaps and listing missing documents. Anything that interprets risk, capacity for loss or suitability belongs to the adviser, within your regulator's rules. Write that boundary into the prompt, and have the adviser review and sign off every draft note before it joins the file.
What if a client does not tick the AI consent box?
The automation should branch. With consent, the AI drafts the file note. Without it, the Zap skips that step and creates a task for a person to write the note by hand. Everything else in onboarding carries on, so declining costs the client nothing and costs you a little time.
Should ID documents pass through Zapier or the AI step?
Keep them out. Let the client upload identity documents directly to your identity-check provider, and pass only a status and a link into the CRM. Fewer copies of sensitive documents means fewer places to secure, and nothing in the file note needs a passport number or an account number.
How do I know a Zap has silently stopped working?
Build two safeguards. Switch on failure notifications to a monitored inbox, and run a weekly count comparing form submissions with CRM contacts created. Zapier's help centre says it turns a Zap off when it errors on 95 per cent of runs and has run more than 20 times in seven days, so silence can mean it has already stopped.
Further reads
- Automate Client Onboarding With AI: Forms, Contracts, and Emails — Forms, contracts and emails for onboarding in any small business.
- AI Governance Checklist for Small Financial Advice Firms — A governance checklist for small financial advice firms.
- AI Agents for Financial Advisers: What They Can Do Without Advising — What AI agents can do for an adviser without advising.
- Jump AI Set-Up for a Three-Adviser Firm: Templates, Consent, Review — Consent and review for AI meeting notes in a small firm.
- Jump AI With Wealthbox: Syncing Notes, Tasks and Contacts — How notes, tasks and contacts sync between Jump and Wealthbox.
- AI for Wealth Management Firms: A 90-Day Rollout Plan for Small Firms — A wider 90-day rollout plan for AI in a small firm.
- Best AI App for Wealth Management: Jump, Zocks and Wealthbox — Jump, Zocks and the Wealthbox notetaker compared on price, recording and archiving, with a break-even sum and a pilot a small firm can run.
- Claude for Financial Advisers: Plugin, Connectors and Plan Needed — What the Claude for Financial Advisors plugin does, which connectors come with it, why Anthropic recommends Enterprise for regulated firms, and the cost.
- Holistiplan Pricing: Basic vs Premium Tax and Who Needs Which — Holistiplan Basic Tax vs Premium Tax priced by household tier from 30 to 750, with a Premium break-even, the billing rules and a 7-day trial checklist.
- HubSpot for Financial Advisers: Pipeline and Where Breeze Stops — Prospect and annual-review pipelines in HubSpot, what email logging does and does not keep, and where Breeze AI needs limits in an advice firm.
- AI Chatbot for Finance Firms: Client Questions Without Advice Risk — What a finance firm's website chatbot may answer, what must go to an adviser, and the disclosure, logging and sign-off a compliance lead will expect.
- AI Chatbot for an Adviser's Website: FAQs and Bookings, Not Advice — Scope, refusal and escalation scripts for an adviser's website chatbot, plus a test pack to run before it goes live.
- AI Website Copy for Financial Advisers: Service Pages and Bios — Build useful adviser pages from verified facts and give each page its own compliance evidence trail.
- Agency Client Onboarding With AI: Questionnaire, Access and Kick-Off — One onboarding flow for an agency: an AI-tailored questionnaire, a safe access checklist and a kick-off agenda drafted from the answers.
- Copilot for Financial Advisers: Outlook, Teams and Excel With Records — Copilot tasks for financial advisers in Outlook, Teams and Excel, sorted by whether the record stays compliant, with current per-user costs.
- FINRA AI Guidance for Advisers: A Checklist for AI-Written Content — What FINRA's generative AI notice means for client emails and posts, turned into a sign-off checklist a small firm can run before anything goes out.
- AI Tools and AI Development: The Complete 2026 Guide — the AI hub, including every tutorial in the AI-for-business series.
Sources: Wealthbox pricing and integrations pages, Zapier's Wealthbox integration page, RightCapital integrations page, Zapier pricing page and help articles on AI model tiers, task limits, Zap shut-off and Human in the Loop, Make pricing page and Human in the Loop app page, Jotform pricing page, read between 27 October and 4 November 2026.