Financial planning data gathering that turns a household's shoebox into cited plan inputs in five days
Plan inputs ready in five days, not three weeks, every line tied to its source page.
See one case, screen by screen ↓Financial planning data gathering, also called client data gathering or financial planning data collection, is the work of collecting a household's statements, policies, pay stubs, tax return and questionnaire and turning them into the planning software's inputs before the plan is built. Financial Planning Data Gathering gives that work to seven agents that read every document into a household balance sheet with each value cited to its page, merge duplicate accounts, flag stale statements and ask the client only for what is missing. A paraplanner reviews and signs, and a financial planner approves before the inputs are pushed.
AI in asset management: a buyer's guide to agentic AI for investment managers →
Updated
Before an advisor can build a financial plan, someone has to do the financial planning data gathering: turn a household's shoebox into the planning software's fields — brokerage and retirement statements, bank statements, 529 and HSA statements, an equity award summary, insurance policies, pay stubs, a mortgage note, a tax return and the client questionnaire. A paraplanner re-keys the balances by hand, works out which statement is the same account, and sends a generic request list that asks the client for things already on file.
The risk sits in what nobody had time to check. A 403(b) balance that is nearly four months old, the same IRA entered from the statement and again from the questionnaire, an insurance cash value typed as the face amount, a foreign contract left out or converted at an unknown rate. Account aggregation covers the linked accounts and misses the policies, equity awards and foreign contracts. And nothing records where a number came from, which is a Reg BI care-obligation and documentation weakness.
Estimated split of the demo firm's 16-day baseline by hand, and of the 5-day target with the agents.
Seven agents sort, read, merge and check every document a household sends; a paraplanner signs and a planner approves the push.
Priya and Sam Natarajan of Scarsdale, NY met their planner, Marcus Webb, CFP, at Quillfield Private Wealth on Wed Oct 14. Their plan meeting is Oct 28, and the plan inputs are due Wed Oct 21 under the firm's 5-day rule. Here is how their financial planning data gathering goes, in the working solution.
Lena has 14 households, 5 ready for her review, 3 waiting on clients and 4 at risk of the 5-day rule. Overnight the Intake agent sorted 61 new documents for 9 households, 47 from the client portal and 14 from the plan mailbox. First in her list: the Natarajans, where 4 lines need her — 2 stale statements, 1 merge question and 1 disagreement between agents.
Matching agent: “Natarajan: 41 mentions → 17 accounts. Kept the two 529 plans apart by beneficiary; one IRA merge is a question for you.”
The Natarajans sent 23 documents between Oct 15 and Oct 18: 19 through the client portal and 4 by email to the plan mailbox, one of them in French. Lena can replay the run: each document gains its type, institution and date, the Extraction agent reads 112 values with their pages, the Matching agent turns 41 mentions into 17 accounts, and the Checker agent re-reads and checks the dates.
Each agent's step is shown as it runs: “Checker agent · re-reading and checking dates — 2 statements older than 90 days · 1 disagreement.”
Financial assets $3,080,826.70, held away $1,609,813.51, liabilities $877,433.64, net worth $2,203,393.06 excluding the home. Each line carries its source page, its confidence and chips for held away, stale or merged. Eighteen lines read at 95% or above can be accepted in one step; the 4 that need Lena stay open in the thread beside the balance sheet.
“4 lines need you · 18 lines at 95% or above are ready to accept.”
Lena types the question into the thread. The answer cites the page: Sam's 401(k) at Harlow Ridge Retirement Services is $512,000.38 as of Sep 30, 2026, “Your account balance” on the quarterly statement. The questionnaire said “about $450K”; the statement is newer and wins under the firm's source-precedence rule. It is a held-away account, so it is not in the custodian feed.
Every answer cites its page, and the agents can act on an instruction — re-read a page, flag a line.
The 401(k) appears four times — the statement, the beneficiary confirmation, a pay stub and the questionnaire. They are merged because the plan name, employer and last four digits agree and the pay-stub contributions match the statement's $1,100.00 contribution line. The two Empire Scholars 529 plans, Anika's $68,250.90 and Rohan's $41,108.33, are kept apart because the beneficiaries differ. The questionnaire's “old IRA ~ $90K” matches only the Bramwell rollover IRA of $94,377.12, at confidence 0.88, so Lena confirms it.
The Matching agent never merges different owners or beneficiaries, and asks below 0.90.
The whole-life line opens the Northfield Mutual Life annual statement of Jul 12, 2026 at page 3, with the value highlighted in the “Policy values on your anniversary” table: cash value $86,240.00, guaranteed $68,410.00 plus paid-up additions $17,830.00, no policy loan. The death benefit, $1,000,000, stays on the coverage line. The statement is 99 days old and current, because insurance annual statements are accepted for 13 months.
Priya's assurance-vie at Mutuelle Rive Gauche Vie was read in French — “Valeur de rachat totale 148 300,00 €” — and shown beside the English. It is converted at EUR/USD 1.1612 (Oct 16, 2026) to $172,205.96. It is 292 days old, past the 90-day rule, and so is her 403(b) at Corbel Benefits Group, $238,914.06 as of Jun 30 — 111 days. Both go on the client checklist.
“Statement is 292 days old — rule ST-01 asks for one under 90 days.”
The Checklist agent drafted a four-item request: the Sep 30 403(b) statement, a 2026 valuation of the assurance-vie, a property tax bill or appraisal for the home, and wills or trust documents if any. It skipped 19 items a generic checklist would ask, because they are already in the shoebox. Tomás edited the message; Lena approves it before it reaches the client. Reminders go on days 3 and 6, with escalation to Marcus on day 7.
The Checklist agent never sends to a client — a second person approves every message.
The Checker agent disputed the auto loan: principal $28,902.14 against a payoff of $29,047.66 on the same page. The plan field takes principal, so Lena keeps $28,902.14 with a reason code. She confirms the IRA merge, marks the home value as the client's estimate and notes the stale statements as asked of the client. The push preview shows the 58 fields that will change in the planning software — 52 to create, 6 to update — and she signs.
Marcus approves with the reason “Inputs traced to sources”. Lena signed the review, so she cannot approve its push. The Plan mapping agent pushes the 58 fields in one step, reads every one back from the planning software — 58 of 58 match — and files the receipt, push PS-2026-118204, on the audit trail. The inputs were due Wed Oct 21; the deadline is met.
“I approve what I can trace. If a number has no page, it doesn't go in my plan.” — Marcus Webb
The same household, seen by five of the people who carry it — what their week looked like, and what it looks like now.
Seven agents sort, read, merge and check every document a household sends; a paraplanner signs and a planner approves the push.
Treats each household as one unit of work: runs the other agents in order, watches the 5-day clock and posts every finding to the household thread, with its citation.
Picks up client-portal uploads, plan-mailbox email and document-vault files; splits combined PDFs; classifies type, institution, owner and statement date; matches each to its household.
Reads balances, holdings, contributions, beneficiaries, policy values, loan terms and salaries — each with page, region and confidence — in the document's own language, and converts foreign currency at a dated rate.
Merges mentions of the same account across statements, pay stubs, the tax return, the questionnaire and the feeds, applies the firm's source precedence, and says why it merged or kept apart.
Re-reads every value independently, ties it to statement totals and feeds, applies the staleness rules and disputes the Extraction agent when they disagree, showing both readings.
Drafts a short client request for only what is missing, stale or estimated, tracks it and re-runs the household when the client uploads.
Maps each approved line to the planning software's fields, builds the field-level push preview, pushes after approval and reads every field back.
Maker, checker, evidence. The agents propose; a named person decides.
Ask about any household, line or document — or tell the agents what to do. Every answer cites its page.
Why is the 401(k) $512,000?
Sam's 401(k) at Harlow Ridge is $512,000.38 as of Sep 30, 2026 — “Your account balance” on the quarterly statement, page 1. The questionnaire said “about $450K”; the statement is newer and wins under rule SP-01. It is held away, so it is not in the custodian feed.
Which statements are too old?
Two are past the 90-day rule (ST-01): Priya's 403(b), $238,914.06 as of Jun 30, 2026 — 111 days; and her assurance-vie, €148,300.00 as of Dec 31, 2025 — 292 days. Both are on the client checklist. The whole-life statement is 99 days old, but insurance statements are current for 13 months (ST-02).
Where does the whole-life cash value come from?
$86,240.00, from the “Policy values on your anniversary” table on page 3 of the Northfield Mutual Life statement — guaranteed $68,410.00 plus paid-up additions $17,830.00; no policy loan. The death benefit is $1,000,000, on page 1.
What is held away?
11 lines, $1,609,813.51 in all: the 401(k) $512,000.38, 403(b) $238,914.06, vested RSUs $214,560.00, assurance-vie $172,205.96, savings $120,400.00, rollover IRA $94,377.12, whole-life cash value $86,240.00, the two 529s $68,250.90 and $41,108.33, checking $42,816.55 and HSA $18,940.21.
13 screens from the working solution, on its sample data. Pick one to see it large.
The rules the agents apply are settings — versioned, and each change approved by a second person.
| Setting | Default | Choose from |
|---|---|---|
| Investment statements are stale after | 90 days | Used by the Checker agent · a change needs a second approver |
| Insurance annual statements are stale after | 13 months | Used by the Checker agent · raised from 12 months in settings v41, with a second approver |
| A value goes to a person below confidence | 0.85 | Used by the Extraction agent · a change needs a second approver |
| A merge becomes a question below | 0.90 | Used by the Matching agent · a change needs a second approver |
| Never merge accounts with different | owners or beneficiaries | Used by the Matching agent · a change needs a second approver |
| Source precedence | custodian feed › statement › aggregation › questionnaire | Used by the Matching agent · a change needs a second approver |
| Foreign currency rate | reference rate, 4:00 p.m. London, prior business day | Used by the Extraction agent · a change needs a second approver |
| A hand edit needs a second reviewer above | $250,000 | Used by the Household coordinator · a change needs a second approver |
Every figure is labelled: a target the solution is built to, an estimate, a typical published result, or a proven one.
“demo” = seen in the working solution, on its sample household data · “target” = the design goal, measured in the live solution; 16 days is the demo firm's 2025 baseline · “estimated” = our estimate · “typical” = published figures (Kitces Research: a 10-hour median to create and deliver a comprehensive plan, and about a quarter of planners deliver it within two weeks of the discovery meeting, from an FPA study; Morningstar Investor Perspectives Advisor study, Oct 2026: 26% would most like to remove paperwork). SEC Rule 204-2 requires books and records for 5 years. Quillfield Private Wealth and the people, households, institutions and account numbers named on this page are fictional — characters and sample data in the working solution.
Financial planning data gathering, also called client data gathering or financial planning data collection, is the work of collecting a household's statements, policies, pay stubs, tax return and questionnaire and turning them into the planning software's inputs before the plan is built. Financial Planning Data Gathering gives that work to seven agents that read every document into a household balance sheet with each value cited to its page, merge duplicate accounts, flag stale statements and ask the client only for what is missing. A paraplanner reviews and signs, and a financial planner approves before the inputs are pushed.
It reads brokerage, retirement, bank, 529, HSA, equity award, insurance and loan statements, pay stubs, tax returns and the client questionnaire, in any format and language. They arrive through the client portal, by email to the plan mailbox or from the document vault, and the Intake agent splits combined PDFs and matches each one to its household. A password-protected file or an unreadable phone photo is shown to a person with a one-click ask to the client.
Held-away assets are accounts the firm does not custody, so they are not in the custodian feed. The agents read them from the client's own statements and from account aggregation, and every held-away line on the balance sheet carries its own dated, cited balance. For the Natarajans that was 11 lines, $1,609,813.51, including a 401(k), a 403(b), vested RSUs, a whole-life cash value and a French savings contract.
Every line carries its as-of date. The Checker agent applies the firm's staleness rules — 90 days for investment statements, 13 months for insurance annual statements in the demo firm's settings — and a stale statement goes on the client checklist. Foreign-language documents are read in their own language, shown beside the English and converted at a dated rate under the firm's currency rule.
It keeps the questionnaire, the firm's data gathering form, as one source among many and replaces the generic request list. The Checklist agent drafts a short checklist of only what is missing, stale or estimated — four items for the Natarajans — and skips what is already in the shoebox. A person approves every message before it reaches the client, and the household is re-run when the client uploads.
It records where every plan input came from: the document, page and region of each value, the reason code on every edit and approval, and who signed and who approved. The audit trail is hash-chained and kept for 6 years, and compliance can export a regulator pack per household with the inputs, sources, approvals, push receipt and read-back.
Yes. The agents read, merge, check and propose; they never push, never send to a client and never approve. The paraplanner settles each open line with a reason and signs, a different person approves the push, and every field pushed to the planning software is read back.
The Agentic Solution Engine builds and deploys it from your requirements and documents — your data rules, your planning software field map and a sample of past household documents — and it goes live once every quality gate has passed. We will walk you through it on your own households first.
Accounts open the same day, and not-in-good-order stops before the custodian sees it.
1-min demoSource of Wealth VerificationEvery dollar of a client's wealth traced to cited evidence before the BSA/AML Officer signs.
1-min demoReg BI Compliance ReviewEvery recommendation reviewed overnight; every flag decided by a principal, with evidence.
1-min demoWe'll run Financial Planning Data Gathering on a sample of your own household documents.