The 30-Minute AI Review Prep Workflow

If you run a registered investment advisory on the North Shore, quarterly and annual client reviews are probably eating more of your week than they should. The fix is not new portfolio software. It's a 30-minute prep routine built around a tool a lot of advisors already pay for: Claude.

A repeatable AI-assisted review workflow turns scattered account data, meeting notes, and market context into a client-ready summary in about 30 minutes, with Claude handling the drafting and organizing while the advisor checks every number and keeps final judgment calls in human hands.

Why Client Review Prep Eats Your Week

Review season is the quiet tax on running a registered investment advisory. If you manage even 80 or 100 households across Lake Forest, Winnetka, and Glencoe, pulling performance data from your custodian, checking for cost basis changes, and drafting talking points for each meeting can burn 2 or 3 full working days a month. None of that work is billable. All of it is necessary.

Peter Drucker put it plainly: efficiency is doing better what is already being done. Review prep is rarely the wrong task. It's just done the slow way, by hand, every single quarter.

The instinct is to buy new portfolio management software. That's usually the wrong fix. Most RIAs already run Wealthbox, Redtail, or Orion for the data, and Salesforce or HubSpot for the relationship history. The gap isn't data. It's the hour spent turning that data into something you can say out loud in a meeting.

What Can Claude Actually Do Before a Review Meeting?

Claude is a large language model built by Anthropic that can read long documents, summarize them, and draft written material in a specific voice. That's the one-sentence version. For review prep, the useful part is the context window: according to Anthropic's Claude documentation, the model can hold up to 200,000 tokens in a single conversation, roughly a full set of quarterly statements, cost basis reports, and your own meeting notes for one household at once.

It cannot log into your custodian and pull live positions on its own. It cannot make a trade or a recommendation you are willing to stand behind without checking it first. What it can do is take the exports you already have, in PDF or CSV form, and turn them into a clean draft: a performance summary, a list of account changes since the last review, and a set of talking points in your own voice.

The economics matter too. A flat Claude subscription charges one monthly price regardless of how many client files you run through it that week, unlike metered API access billed per token. For a solo advisor or a small team doing this every quarter, that flat structure is the reason the workflow below is worth setting up once and reusing.

If you are not sure which of your current tools already cover this and where the actual gap sits, run our free AI readiness quiz before you build anything new.

The 30-Minute Workflow, Step by Step

This is the version worth setting up for a solo advisor or a two-person team prepping for a Tuesday morning review. It assumes you already export account data from your custodian or portfolio system, something almost every platform supports.

Setting this up once, as a saved prompt and a folder structure, is the kind of work our AI consulting engagements focus on for North Shore advisory teams: not buying new software, but wiring up the tools you already pay for.

Is It Compliant to Use AI on SEC-Regulated Client Communications?

A registered investment adviser is a firm or individual registered with the SEC or a state regulator that provides investment advice and owes its clients a fiduciary duty. That duty does not disappear because a draft came from Claude instead of an associate. It just moves the review responsibility to you.

The workflow above is built to keep a human decision at every point that matters. Claude drafts the summary. It does not decide what belongs in it, and it never sees your custodian credentials, because you are uploading exported files, not connecting an account. Treat every draft the way you would treat a summary written by a new associate: useful, fast, and not client-ready until you have checked it line by line.

Keep your compliance officer looped in on the workflow itself, not just the output. Most firms already have a policy for associate-drafted communications. Extending it to cover AI-assisted drafts is a smaller lift than writing one from scratch.

Where This Breaks Down, and What to Do Instead

This workflow is not a fit for every review. A household with a complicated trust structure, an active options overlay, or a recent liquidity event needs your full attention from the start, not a draft to edit. Use the 30-minute version for the standard quarterly check-in and save your full attention for the accounts that actually need it.

It also will not work if your data exports are inconsistent: a PDF that doesn't parse cleanly, a CSV missing a column Claude needs to compare periods. Fix the export first. A workflow built on messy inputs produces a messy draft no matter how good the model is.

A client review is exactly that kind of hosting. The prep work is not the point. What you do with the 30 minutes you get back, whether that's a longer conversation about a client's actual goals or just getting home for dinner, is the point.

I saw a version of this discipline building fraud-detection AI used by Blue Cross Blue Shield across 4 states: the model does the pattern work, a person owns the decision. The same split holds here. See how it plays out for a North Shore family in our family office case study, and browse more workflows like this one on the Bace Agency blog.

Set up the folder structure and the saved prompt this week, before next quarter's reviews start piling up. The workflow gets easier every time you run it.

If you want help wiring this into your actual CRM and custodian exports instead of building it yourself, a free 30-minute AI audit will get you a one-page plan specific to your book of business.

Frequently Asked Questions