# AI in Wealth Management: Meeting Notes and Client Onboarding

> An AI assistant can write up a client meeting well and still file a client's health detail in the client-records system or a maybe as a plan. Where each draft lands, who approves it there, and what the hours after every meeting cost.

By Vlad Shvets, Partner @ Qvery · Published 2026-10-30 · Updated 2026-11-09 · 8 min read
Canonical: https://qvery.ai/blog/ai-in-wealth-management/

At an advisory firm of about 150 people, every client meeting ends in paperwork: the file note, the firm's written record of what was advised and decided, the email to the clients restating what happens next, the changes to the CRM, the client-records system the firm works in, and the tasks for whoever handles the forms.

Advisory teams spend more than an hour preparing for and following up on every hour with clients, and the email the advisor sends is a record the firm keeps for five years, wrong line included.

That ratio comes from Kitces Research's 2024 survey of 621 advisory teams, mostly small independent ones.

> 🗂️ At Qvery, we help advisory firms give their advisors back the hours that go on file notes and follow-ups and keep the service team off retyping, with custom AI agents that draft each meeting's note, email and CRM changes for the advisor to approve and check every new-account packet against one source.

## What Follows Every Client Meeting

**Client Meetings Take 16% of an Advisory Team's Time per Client** (share of team time per client)

Meetings take 16% of an advisory team's time per client; the preparation and follow-up around them take 23%.

| Label | Value |
|---|---:|
| Client meetings | 16% |
| Client correspondence | 16% |
| Servicing requests | 15% |
| Meeting prep | 13% |
| Recurring planning | 11% |
| Meeting follow-up | 10% |
| Client check-ins | 10% |
| Ad hoc planning | 8% |

Share of a team's ongoing time per client after the first year: 621 US advisory teams, mostly small independent firms, surveyed August to September 2024. Source: Kitces Research: The Kitces Report Vol. 2, 2024 Financial Planner Productivity Study, Figure 4.20

Run it on your own book. At 21 hours a client a year, 10% of it meeting follow-up, each household costs **about two hours** a year of follow-up, and a firm serving 2,000 households around 4,200 hours, before the client correspondence Kitces counts separately.

Personal financial advisors earned a mean $156,670 in May 2025, the Bureau of Labor Statistics reports; with benefits at 30% of private employer costs, that is about $223,800 a person, so twenty advisors giving 10% of a 2,080-hour year to follow-up, about the same hours, cost about $447,600 a year.

(An interactive calculator on the page: roi-advisor-meeting-follow-up.)

With AI agents drafting 65% of it, our assumption, the cost falls to $168,660 a year ($156,660 of advisor time left and $12,000 to run them): $278,940 and 2,704 hours of advisor time handed back, and every $10,000 of the build earned back in 0.4 months.

Writing up a meeting is the kind of work we build agents for: it repeats after every meeting, it draws on one transcript and one CRM, and a person can check the result against what was said.

AI agents for finance: https://qvery.ai/finance/

## What the Agents Draft

An invented annual review with a married couple holds what a real one does, and each item has to land in a different place:

- An inherited IRA (an individual retirement account), and a backup beneficiary to change after the client's mother died.
- A husband's stent, and his "maybe sixty-two" on retirement.
- A kitchen the couple "might" fund next spring, waiting on a contractor's quote.
- A transfer into a Roth IRA they asked for outright: $500 a month from joint checking, starting in January.
- A balance in a 401(k), the workplace retirement plan, that the client corrected mid-sentence, from $320,000 to $340,000.

In our own test, given a transcript like that and the request an advisor types first, general AI assistants heard the meeting well and put things in the wrong place: a possibility filed as a planned withdrawal, a client's health detail written into the CRM.

The agents we build for your advisors keep those apart. When the notetaker's transcript lands, they draft the file note, the email, the tasks and the proposed CRM changes for the advisor who held the meeting, with each possibility kept a possibility and health details kept out of the CRM and the email.

A new client's paperwork is the second job. Someone on the service team retypes the signed advisory agreement and the transfer forms into the CRM, the planning software and the forms of the custodian, the firm that holds the clients' assets, and every retyped field can get a form sent back for errors.

The agents read the whole packet as each document arrives, check every field against one source, and hand back the exceptions: what is missing or mismatched, the forms filled from the source for review, and the next request to the client. When two documents disagree, they list the conflict and never pick a value.

## What a Person Approves

![Four steps: the agents check the packet, operations verifies the client and sends it, the asset holder accepts it or returns a form, the transfer runs; a returned form loops back to the check.](https://qvery.ai/blog/ai-in-wealth-management/illustration-onboarding-loop.webp)

*The agents' work is the check and the chasing; the submission and the identity check stay with people.*

**What the agents may do with each draft, and who approves it**

| Where it lands | What the agents may do | Who approves it |
|---|---|---|
| File note | Draft what was advised and decided; flag a sensitive detail, never write it out | The advisor who held the meeting |
| Client email | Draft the instructions and promises; a possibility reads as a topic to discuss | The advisor |
| Client-record change | Propose only the fields that changed; a possibility never becomes a plan | The advisor or a client service associate |
| Client instruction | Nothing: transfers, beneficiary and address changes need the client's signed form | Operations |
| New-account packet | Check it against one source and list the exceptions | Operations, which verifies the client and submits it |

Approving one approves none of the others: signing off the file note sends no email, changes no profile and starts no Roth transfer, which still waits for the client's signature.

A person verifies the client's identity and submits the packet; the agents check the packet, not the person.

## Bring Us a Month of Recorded Reviews

Bring a month of review meetings your clients agreed to record. The agents draft them beside what your advisors wrote, the advisors approve each draft, and your compliance officer owns the recording rule; a first working version runs on one of the two jobs within two weeks once access is in place.

What the agents may read and do is set in a written agreement, your data is processed where that agreement says, and every action the agents take is logged.

Have Qvery Build Your Advisory Paperwork Agents: https://calendly.com/qvery-vlad/30min

## Frequently asked questions



### Is AI going to take over wealth management?

It can take on the paperwork around the advice: the file note, the follow-up email, the CRM changes and the account-form checks, each reviewed by a person. The advice, the client's decisions and every approval stay with people.

### Should an AI notetaker write straight into our CRM?

No. Every CRM change should arrive marked proposed and wait for the advisor or a client service associate to approve it.

### Do we have to tell clients that AI writes the notes?

That is a disclosure decision, separate from consent to record.

### Can AI agents coordinate a new client's onboarding across systems?

They can read the agreement, the intake data and the custodian's forms, check that they agree, fill the forms from one source and list what is missing; where the custodian's digital forms already check the input, their job shrinks to the exceptions and the chasing. Operations still submits the packet, the custodian decides whether it is complete, and people verify the client's identity.

### Can AI agents verify a new client's identity?

No. The agents check that the packet agrees with itself; a person verifies the client.

### What does Qvery build for an advisory firm?

Custom agents for your firm that draft each meeting's note, email, tasks and proposed client-record changes for the advisor, and check new-account packets, in the systems your team already uses. A first version runs within two weeks once access is in place, we train its approvers, and each week we count hours saved and errors avoided.

### How much time do financial advisors spend on meeting follow-up?

Advisory teams spend more than an hour preparing for and following up on every hour with clients, in Kitces Research's 2024 survey of 621 teams, mostly small independent firms. Client meetings took 16% of a team's ongoing time per client after the first year.

### How many hours of follow-up does an advisory firm carry a year?

At 21 hours a client a year, 10% of it meeting follow-up, each household costs about two hours of follow-up a year. A firm serving 2,000 households carries around 4,200 hours, before other client correspondence.

### What does an advisor's hour of paperwork cost the firm?

Personal financial advisors earned a mean $156,670 in May 2025, the Bureau of Labor Statistics reports. With benefits at 30% of private employer costs, each costs about $223,800 a year.

### How much can AI agents save an advisory firm on meeting follow-up?

In the post's example, twenty advisors spend 10% of a 2,080-hour year on follow-up. Assuming agents draft 65% of that work, annual costs fall from about $447,600 to $168,660. That returns $278,940 and 2,704 hours of advisor time.

### Why does each item from a client review belong in a different place?

Because each means something different: a backup beneficiary to change on an inherited IRA is a task, a client's stent is a health detail to keep out of the CRM, and a kitchen the couple might fund next spring is a possibility, not a planned withdrawal.

### What do general AI assistants get wrong in meeting notes?

In our own test, they heard the meeting well and filed things in the wrong place: a possibility recorded as a planned withdrawal, a client's health detail written into the CRM.

### How do AI agents handle a client's health details?

They flag a sensitive detail in the file note and never write it out, and they keep it out of the CRM. The advisor who held the meeting decides what the record says.

### Does approving the file note send the client email?

No. Approving one draft approves none of the others: signing off the file note sends no email, changes no profile and starts no transfer, which still waits for the client's signature.

### How is access to advisory client data controlled for AI agents?

For your advisory firm's paperwork, a written agreement sets what the agents may read and do. Your data is processed where that agreement says. Every action the agents take is logged.

### Why does a wrong line in a client email matter?

The follow-up email an advisor sends is a record the firm keeps for five years, wrong line included. So each draft waits for the advisor who held the meeting to approve it before it goes.

### Who owns the rule on recording client meetings?

Your compliance officer. The agents start from a month of review meetings your clients agreed to record, drafted beside what your advisors wrote, and the advisors approve each draft.

### What does it cost to run meeting follow-up agents?

The post's example allocates $12,000 a year to running agents for twenty advisors. It assumes 10% of a 2,080-hour year goes to follow-up and agents draft 65% of that work. The remaining advisor time costs $156,660, bringing annual costs to $168,660.
