Claude for financial advisors: what Anthropic launched, and what a solo advisor can use today
A review at ten, a client who just changed jobs at two, a fund document to explain by Friday. Claude can take real work off that day, and Anthropic now offers an edition built for advisory firms. Here is what it launched on September 14, 2026, who it is for, what a solo advisor can use today, what to keep out of the chat, prompts that fit a practice, and the part of review prep it still cannot do: remember what the client sent you.
Short answer: Claude for Financial Advisors is Anthropic's edition for advisory firms, released on September 14, 2026 as "a suite of connectors and workflow skills" that plug Claude into a firm's custodian, portfolio, planning and CRM systems. Anthropic recommends its Enterprise plan for a registered investment adviser. A solo advisor or wealth manager on an ordinary paid plan gets ordinary Claude, which still does useful work. For the wider picture, including the questions to put to compliance first, see our guide to AI for financial advisors. Every Anthropic statement below was read at the source on 12 October 2026.
This guide is published by Equerry, a personal assistant for client work. Equerry is not affiliated with Anthropic, and nothing on this page is investment, legal or compliance advice. Your firm's policy decides what goes into any tool.
What Anthropic launched on September 14, 2026
Anthropic's announcement opens with "Today we're releasing Claude for Financial Advisors, a suite of connectors and workflow skills", and those two words carry the design. Connectors reach the systems a firm already runs. Anthropic's page names Addepar, BlackRock, Charles Schwab, Envestnet, iCapital, Orion, SS&C Black Diamond, Vanguard, Wealthbox, Wealth.com and Zocks, alongside existing connectors such as Microsoft 365 and Salesforce; its line on Schwab begins "Charles Schwab connects Claude to Schwab Advisor Services custodial data", then lists balances, positions and transactions among what that covers. Skills package the work. There are eight, as Anthropic's announcement and repository list them (its plugin page lists six of them as example uses, with the estate and tax brief shown as two): Advisor onboarding, Alternative investments brief, Compliance and AI policy, Estate and tax brief, Portfolio rebalance review, Post-meeting notes and follow-up, Pre-meeting prep, and Prospect intake.
Anthropic's example shows the shape of it: the advisor types "/pre-meeting the Ostrelbys, Q3 check-in, tomorrow at 10am." and Claude pulls the record, tasks, performance and plan snapshot from the connected systems into "one prep doc you can read in 10 minutes." Afterwards, "Claude turns the meeting into a CRM note, the follow-ups you committed to, and the opportunities it heard," and "You approve once, and Claude writes what you approved to your CRM."
On the line every advisor will ask about, Anthropic is direct: "Claude does not give investment advice." "It pulls the information together and drafts the analysis for your review." "No. Claude doesn't place trades or move money." And "Notes, emails, and CRM updates are drafted and held until you approve them." The compliance skill "screens client-facing language against the SEC Marketing Rule to flag potential issues"; Anthropic's page calls that a pre-check ahead of the firm's own review, not an approval. The open-source repository (Apache License 2.0) puts it in one plain sentence: "Users remain solely responsible for all investment decisions, client communications, supervisory review, recordkeeping, and compliance with applicable laws, regulations, and firm policies."
Who it is for, and what a solo advisor can use today
Availability, from Anthropic's pages: "The advisor plugin is available today." "The plugin works on" "paid Claude plans," and "we recommend Enterprise for Registered Investment Advisor" "as it is the only plan that includes audit logs and compliance controls for a regulated firm." Anyone without one is told, "If you don't have an Enterprise license, request one using this form", and the sentence goes on to promise that someone from Anthropic will reach out. One answer from the solutions page's FAQ belongs here too, because a Pro subscriber will meet it: asked "Does it work with Claude or Claude Cowork?", Anthropic answers "Claude Cowork is now just Claude, so you don't have to choose where to get work done" and "The update is rolling out to Pro and Max plans, with more to follow." What is rolling out there is the Cowork change, not the advisor plugin; the plugin's availability is the line above.
Anthropic's page does not speak to a solo practice directly; its closest line is "Claude comes configured for an independent practice." The value sits in the connections to your custodian, portfolio, planning and CRM systems, which are your firm's and your custodian's to approve. Without them, what you have is Claude, and this is what it costs.
| Plan | Price (US, before tax) | Worth knowing for an advisor |
|---|---|---|
| Free | $0 | Up to five Projects; enough to judge the drafting on non-client material |
| Pro | $20/month billed monthly, or $17/month on the annual plan ($200 up front) | More usage, Projects without the cap of five, "Claude in Chrome and Microsoft 365" |
| Max | From $100/month | "Choose 5x or 20x more usage than Pro" |
| Team | $25 per standard seat billed monthly, or $20 billed annually | "For teams of 2 to 150"; "No model training on your content by default" |
| Enterprise | "US$20/seat/month, billed annually. Usage cost scales with model and task." | The plan Anthropic recommends for an RIA using the advisor edition; "None by default" on model training |
"Every plan has usage limits that reset on a rolling five-hour session window." On Free, Pro and Max the plan comparison lists model training as "Opt-out," a setting you control (steps below).
Where Claude saves an advisor time, and where it falls short
- A planning concept in plain words. Why a Roth conversion can make sense in a low-income year, how a beneficiary designation overrides a will, why rebalancing is not market timing. Claude drafts; you check every fact before it goes to a client.
- Your notes into a structure. A half-page from the last review becomes an agenda, a list of open items, or the three decisions the next meeting has to make.
- Long public documents. A summary prospectus, a fund's annual report, an IRS publication. Ask for page references, then open those pages yourself.
- A second read on your own letter. The note after a market drop: shorter, calmer, with any promise it should not make pointed out. Still your letter.
The shortfalls are as specific. Advice: nothing it says is a recommendation. Figures with a date on them: contribution limits, the age at which required distributions begin, bracket thresholds. These change by year and a model's knowledge has a cut-off, so take them from the IRS page, never from the chat. Between chats: it holds nothing unless memory is on, and memory is topics distilled from your chats, not the household file. Portfolio data: none, unless the firm connects it.
What to paste, and what never to paste, by document type
Start from your firm's policy, or the written one you gave yourself as your own chief compliance officer. Within it, sort by document, because each kind fails in a different way.
| Document | Usually workable | Keep out |
|---|---|---|
| Custodian and brokerage statements | Almost nothing. Retype the one question the statement raised. | Account numbers, balances beside a name, a holdings list that identifies a household on its own |
| Plan output from planning software | One scenario, round figures, no names, to draft the explanation | The full report: names, dates of birth, Social Security numbers, beneficiaries, goals written in the client's own words |
| Client emails | The question, retyped, with the person taken out | The email itself: headers, the family or health detail that sits in the same paragraph, the attachments |
| Your meeting notes | Agenda items and open tasks, initials at most | Health, a divorce, an inheritance, a named beneficiary, anything the client asked you to keep between you |
| Prospectuses, fact sheets, IRS publications, public filings | All of it | Nothing, unless your firm marked a copy internal |
Don't over-read Anthropic's line that "Some information is never saved to memory, even if you ask," with "financial account numbers" on the list. That governs memory. The chat keeps what you pasted until you delete it.
Claude's own data settings, step by step
Four controls, read 12 October 2026 at Anthropic's help and privacy centers. They do not answer the compliance question for you.
- Decide the training setting first. On the web or in the mobile app: "Select your name from your settings menu," then "Select Settings," then "Select Privacy," and under "Help Improve our AI models" toggle it off (Privacy Center article dated August 3, 2026). Off, "we will not use any new chats and coding sessions you have with Claude for future model training." Left on, Anthropic "may retain your data in a de-identified format for up to 5 years in our model training pipelines."
- One Project per household. "Projects allow you to create self-contained workspaces with their own chat histories and knowledge bases," and "Each project has its own separate memory space and dedicated project summary." Name it by household and upload nothing the table above rules out. "Free users can create a maximum of five projects."
- For a one-off question, keep memory out of it. In a new chat, before the first message, "Click '+' in the message box," then "Click 'Memory' to turn it off for this chat." Outside a Project you can go further: "Click the ghost icon to enable incognito mode." "Incognito chats are not used for training" and are kept "for either 30 days (default)."
- Read what it kept, and delete in the right order when a client leaves. Under Settings > Memory, "Everything Claude remembers is listed under Topics"; open one and "use the edit icon to change it or select 'Delete' to remove it." A deleted chat is "Removed from your chat history immediately" and "Deleted from our back-end storage systems within 30 days," but delete the household's topics too, because "When a conversation expires or is deleted, related memory entries generated from it won't be removed."
Prompts for client work
Each works from material you supply, with the table's right-hand column kept out. Read everything before it reaches a client.
A review agenda from your own notes
"From my notes since this household's last review (first names only, no figures), draft a 60-minute review agenda: what changed in their life, what I said I would look into, the questions they are likely to ask, and the three decisions we need to make. Leave the numbers to the planning report. Anything my notes do not settle, list under a heading 'still to confirm' and leave blank; invent nothing. Notes: [paste]."
A required-distribution timeline, with the figure left blank
"Explain in plain words, under 200 words, how required minimum distributions work for someone approaching them: when they start, how the first year differs from later ones, and what happens if one is missed. Do not state the starting age, any percentage or any dollar figure. Write [CHECK CURRENT IRS RULE] wherever one belongs, so I fill it in from the IRS page before this goes to a client."
Questions for a client who mentioned a job change
"A client mentioned in passing that they are changing employers next month. Write the questions I should ask before our call, grouped under the old employer's retirement plan, any equity or deferred compensation, insurance and benefit gaps between the two jobs, cash flow during the transition, and timing. Ask, don't advise. Use no names."
A plain summary of a published fund document
"Below is a fund's published summary prospectus. In 150 words, for a client who never reads these, summarise the stated objective, the principal risks the document itself names, and the fee table as written. After each point, give the page number and the first five words of the sentence it rests on. If the document does not say it, leave it out. [paste]."
None of these knows what the client wrote to you in March.
What Claude does not hold: what each client sent you across the year
Anthropic's help center: "Claude saves memory as a set of individual topics as you chat." That is good for remembering how you like a client letter written, not for being the record of a household. On the Enterprise edition, the connectors bring in the CRM record and the custodial data. On Pro, the client's email reaches Claude when you paste it, or when you connect the mailbox it sits in: the whole account, not the one thread. And the things that decide a review are said in passing: the son's tuition deposit due in May, the offer from a competitor, "don't mention the inheritance in front of my sister." Those reach no tool until you put them somewhere. Our wider guide for advisors has the review-meeting routine and the compliance questions that go with any tool. A two-minute note after each call closes most of the gap; the weak point is finding it in October, after a later reply changed it.
Where Equerry fits
Equerry is a personal assistant on your iPhone who helps you get things done. For an advisor it keeps what each household sent and said, from what you give it; Claude keeps the drafting. Plainly: Equerry does not join or record calls, has no CRM or custodian connection, never asks for your inbox, and gives no investment advice.
| What you send | The email in which a client mentioned, two paragraphs down, that they are changing employers in November, forwarded to your Equerry's own email address; the thread with their CPA about the Roth conversion, with Equerry on the CC line; the summary prospectus for the fund their new plan defaults to, shared as a PDF from your iPhone. |
|---|---|
| What you ask | "Before Thursday's call with the Okonkwos: what did they tell me about the job change, and where did their CPA land on the Roth timing?" |
| What you get back | The start date and the line about the unvested match, from the forwarded email; the CPA's view, from the thread; and, if you ask, the fund's expense ratio as the prospectus states it, instead of twenty minutes in your sent folder. When it draws on a note you saved, it carries a source chip: tap it to see the passage Equerry used and open the original. For a thread you CC'd or forwarded: What Equerry remembered from it is listed in Settings → Your memory, each line marked as from that email, and you can delete any line. If you allow calendar access and set up Meeting Prep, a short brief arrives before Thursday's call from the same material. |
The split, prompt by prompt, is the one this page has been drawing: Claude drafts from what you paste; Equerry keeps what each household sent and said, so the paste has something behind it.
| The job | Claude drafts | Equerry keeps |
|---|---|---|
| The review agenda | The 60-minute agenda, from the notes you paste | The notes themselves: the voice note you recorded in Equerry in the car after the spring review, the typed note from the May call. Ask "what have the Okonkwos raised since the spring review?" and the agenda prompt has its material. |
| The job-change call | The question list, with no names in it | The forwarded email where the client mentioned the move, so the start date and the new plan's name come from the email, not from your memory of it |
| The fund document | The 150-word summary | The prospectus PDF you shared, so in October "what did that fund say its expense ratio was?" is a question with a passage to tap, not a search |
| The Roth conversion | The plain-words explanation of why it can make sense in a low-income year | The CPA thread you CC'd Equerry on (replies that keep its address on the thread keep arriving), so the client note you draft starts from what the CPA actually said |
| The promise | A first draft of the 529 comparison, from the two plans' published disclosures | The date. Ask for a reminder on the 14th to send it, and the push arrives on the day you pick with the words you wrote. |
Honest limits. Equerry knows only what you send it, plus your calendar if you allow it, so a client reply you never forwarded is out of its reach. Its address shows on the CC line to everyone on the thread. It is not an archiving or recordkeeping system and makes no claim about any SEC or FINRA rule. And it is an outside service: what you send is processed to make it searchable, including by the providers named on our AI disclosure page, so the compliance question you would ask about Claude applies here in the same words. What changes is reach: one thread and one note at a time, never a whole mailbox.
Sources, all read 12 October 2026. Anthropic: Claude for Financial Advisors (announcement, September 14, 2026); Claude for financial advisors (solutions page and FAQ); Claude for financial services; anthropics/claude-for-financial-advisors README (Apache License 2.0); Claude pricing (prices exclude applicable tax; "Price and plans are subject to change at Anthropic's discretion"). Claude Help Center: Use Claude's chat search and memory to build on previous context, Use incognito chats, What are projects? Anthropic Privacy Center: How do I change my model improvement privacy settings? (dated August 3, 2026), How long do you store my data? (dated July 1, 2026). Anthropic's pages describe the consumer plans unless they say otherwise; where its current page disagrees with this one, Anthropic's page is right. This guide is general information, not investment, legal or compliance advice.
Frequently asked questions
What is Claude for Financial Advisors?
Anthropic's edition of Claude for advisory firms, released on September 14, 2026. Its announcement calls it "a suite of connectors and workflow skills": the connectors link Claude to systems a firm already runs (Anthropic names Addepar, BlackRock, Charles Schwab, Envestnet, iCapital, Orion, SS&C Black Diamond, Vanguard, Wealthbox, Wealth.com and Zocks), and the skills package eight jobs, from pre-meeting prep and post-meeting notes to a portfolio rebalance review and an estate and tax brief. Anthropic's own page says "Claude does not give investment advice" and that notes, emails and CRM updates "are drafted and held until you approve them." It is aimed at the firm and the firm's plan: Anthropic recommends Enterprise for a registered investment adviser. (claude.com, read 12 October 2026.)
Can a solo advisor use Claude for Financial Advisors?
Anthropic's page says "The plugin works on" "paid Claude plans," then recommends Enterprise for a registered investment adviser "as it is the only plan that includes audit logs and compliance controls for a regulated firm." It does not address a solo practice directly; the nearest line is "Claude comes configured for an independent practice." The practical point is where the value sits: in connections to a custodian, a portfolio system, planning software and a CRM. If you have those systems and your firm (or you, as your own compliance officer) approve the connection, read Anthropic's setup page. Without them, what you have is ordinary Claude on Pro or Max, which still drafts, explains and summarises well. Whether any client information goes into it is your firm's decision, not Anthropic's. (Read 12 October 2026.)
How much does Claude cost for a financial advisor?
From Anthropic's pricing page, read 12 October 2026 (US, before tax): Free is $0 and includes up to five Projects; Pro is $20 a month billed monthly, or $17 a month on the annual plan ($200 billed up front); Max starts at $100 a month for 5x or 20x Pro's usage. The Team plan, sold "for teams of 2 to 150," is $25 per standard seat a month billed monthly or $20 billed annually, with "No model training on your content by default." Enterprise, the plan Anthropic recommends for an RIA using the advisor edition, is listed at "US$20/seat/month, billed annually," and "Usage cost scales with model and task." Anthropic publishes no separate price for the advisor plugin itself. "Price and plans are subject to change at Anthropic's discretion."
Is it OK to paste client statements into Claude?
Treat a statement as the document you don't paste. It carries account numbers, balances and a name on one page, and almost every question you would ask Claude about it can be asked without it ("how do I explain cost basis after a step-up" needs no statement). What Anthropic documents, read 12 October 2026: on Free, Pro and Max you choose whether new chats may train its models; a deleted chat is "Deleted from our back-end storage systems within 30 days"; and "financial account numbers" are among the things "never saved to memory, even if you ask". That last line governs memory, not the chat, where what you paste stays until you delete it. Whether any client document may go in at all is a question for your firm's policy and, for an outside service, its vendor review; the questions to ask are in our guide to AI in an advisory practice.
Does Claude remember my clients between chats?
It can. Anthropic says "Memory is on by default for Free, Pro, and Max plans on the web, Claude Desktop, and Claude Mobile," that "Claude saves memory as a set of individual topics as you chat," and that what it keeps includes "The people and places in your work and life." Everything is listed in Settings → Memory under Topics, where each one can be edited or deleted. On paid plans it can also search your past chats when you ask. But memory is a set of distilled topics, not the household file: it holds only what you typed, uploaded or connected, and "When a conversation expires or is deleted, related memory entries generated from it won't be removed." One Project per household keeps them apart, since "Each project has its own separate memory space." More in does Claude have memory? (Anthropic Help Center, read 12 October 2026.)
Does Claude join or record client meetings?
No. Claude works from what reaches it. The advisor edition's post-meeting skill, per Anthropic's open-source README, turns a "meeting artifact or transcript into a CRM note": the transcript comes from somewhere else, such as a firm-approved note taker (Zocks is among the named connectors). Equerry does not join or record calls either. What it offers instead is a voice note you record in Equerry yourself, after the meeting, with what was decided and what you promised, transcribed and kept so you can ask about it before the next review. If your firm uses a note taker, its policy on recording and client consent is the one that governs; our wider guide for advisors covers those questions.
Last updated 2026-10-12.
All product names, logos, and brands are property of their respective owners. Equerry is not affiliated with, endorsed by, or sponsored by any other product mentioned; comparisons are provided for informational purposes.