Claude for Financial Advisers: Plugin, Connectors and Plan Needed

Claude for Financial Advisors is a plugin that adds adviser workflows (onboarding, meeting prep, portfolio reviews, proposals, compliance checks) and connectors to custodian, planning and CRM tools. It runs on paid Claude plans, but Anthropic recommends Enterprise for registered investment advisers because only that plan has audit logs, and Enterprise needs at least 20 seats.

That 20-seat minimum is the practical catch for a small firm: seat fees alone start at $400 a month, with usage billed separately at API rates. A four-person practice on Team pays $80 a month for the same plugin, so the real question is whether the firm's record-keeping duties need Enterprise's audit logs, and that is one for the compliance officer.

Follow me on Instagram@sagnikteaches

What the plugin is, and what it is not

Anthropic describes the plugin as a bundle of skills that you install from the plugin directory in Claude. It is not a separate model, app or adviser. Skills are packaged instructions plus connections to your systems, so the plugin changes how Claude approaches an adviser's tasks and where it can read from. It was first published on 14 September 2026, and Anthropic's launch post has been updated since.

Connect on LinkedInSagnik Bhattacharya

The limits matter as much as the features. Anthropic's pages say Claude does not give investment advice, place trades or move money, and that drafts such as notes, emails and CRM updates are held until the adviser approves them. The launch post adds that investment recommendations, client communications, compliance determinations and other regulated activities remain subject to human review and approval. The compliance skill flags potential issues in client-facing wording, but it does not make the decision.

Subscribe on YouTube@sagnikteaches

In practice that makes the plugin an assistant for preparation and paperwork, which is the same line the tutorial on AI agents that handle admin without advising draws for AI agents in general. If you are weighing it against a general chat tool, the comparison of Claude and ChatGPT for meeting prep and records is a useful companion.

The eight jobs it is built for

Anthropic's launch post lists the skills. Each one is a task an adviser does weekly, and each has a review point, which is what a firm should plan around.

• Onboarding a new adviser and meeting prep. Walks someone through setup and builds a first-meeting brief. Review point: facts about the client's circumstances.
• Alternative investments summaries. Summarises holdings of this kind for a household. Review point: valuation dates and terms taken from the documents.
• Client-facing language review. Screens text for compliance review and helps document an AI-policy process. Review point: the compliance officer's decision, not the tool's flag.
• Estate-plan and tax checks. Checks estate plans against account titling and beneficiaries, and reviews prior-year taxes. Review point: anything legal or tax-specific goes to a qualified specialist.
• Portfolio drift and concentration. Flags drift and concentrated positions with a draft explanation. Review point: target allocations and the client's actual mandate.
• Meeting transcripts to client summaries. Turns a transcript into a summary, a recap email and CRM tasks. Review point: every commitment and number in the recap.
• Pre-meeting briefs. Pulls holdings, activity and open items together before a meeting. Review point: that the data came from the right system on the right date.
• Prospect intake and handoff memos. Organises prospect information into intake summaries. Review point: facts the prospect has not confirmed.

Firms can use the skills as they come or adapt them to their own workflows, which is where a short house-style document pays off, for instance a template recap email that the skill must follow.

Connectors: which come with the plugin and which you add

A connector lets Claude read from a system you already use. Anthropic's pages describe the plugin as connecting to custodian, portfolio, planning and CRM systems, and the launch post separates two groups.

Anthropic's two pages do not group the connectors identically: the solutions page also names SS&C Black Diamond and Vanguard alongside the bundled nine, so check in the guided setup which ones appear for your account. The launch post adds that some partners offer their own plugins, such as BlackRock for its adviser platform. Two practical checks before you rely on any connector. First, a connector reads from your own account at each vendor, so you need the licence or access level that vendor requires, and some vendors may charge for the connection. Ask each one in writing. Second, work out what the connector can reach: a CRM connector that reads every client note exposes more than one scoped to a single book of clients. Start with the fewest connectors that cover your first two tasks, and add more after review.

Why Anthropic points regulated firms to Enterprise

Anthropic's solutions page says the plugin works on paid Claude plans, and the launch post recommends Enterprise for registered investment advisers (RIAs) because, in Anthropic's words, it "includes the audit logs that support recordkeeping". The solutions page calls Enterprise the only plan with audit logs and compliance controls for a regulated firm. The difference is in the admin features around the plugin, not in the plugin itself.

Here is what the plan table on Anthropic's pricing page shows for the two tiers a small firm would consider.

What the audit logs actually record

Anthropic's help article on audit logs says they are available for Enterprise organisations only, that owners can export them from the data and privacy settings, and that an export covers the past 180 days. Logged events include sign-ins and sign-outs, project creation and deletion, file uploads, user invites and removals, SSO configuration changes, data export starts and conversation creation, renaming and deletion.

The same article says the titles and content of chats and projects are not included in audit logs, only their unique identifiers. If the firm needs the text of prompts and answers, Anthropic's Enterprise documentation says the Compliance API gives programmatic access to activity logs, chat histories and file content. The audit-log article also points to a manual route: the Primary Owner can run an organisation data export that includes chat inputs and outputs, on Team as well as Enterprise, although chats deleted before an export are left out of it. That distinction is crucial for a regulated firm: audit logs prove that an event happened and who triggered it, while the Compliance API is the route to the content. Which record you need depends on your obligations, so it is a question for your compliance officer and, if necessary, your regulator's guidance. Read the help article on accessing audit logs before you decide.

Retention differs too. Anthropic's privacy centre says that deleted conversations leave the back end within 30 days, that content flagged by safety systems may be kept for up to two years, and that custom retention controls sit in Enterprise. Team admins cannot set a firm-wide retention period.

The plan sums: Team against Enterprise for 4, 6 and 12 people

Seat prices on both tiers are $20 a month on annual billing for the standard seat. The difference is the Enterprise minimum, and that Enterprise usage is billed separately. Anthropic's help article says the Enterprise seat fee covers access only and all usage is billed at API rates, with admins able to set spend limits at organisation or user level.

The table shows seat fees only. Enterprise usage is extra and unpredictable until you have measured it, whereas Team's allowance is bundled into the seat. So the real cost of Enterprise for a small firm is the unused seats plus whatever usage the firm generates, which you should budget by running a pilot with spend limits on. On self-serve Enterprise that usage is paid for with credits bought up front, whereas sales-assisted plans bill it monthly in arrears. Sales-assisted Enterprise needs 50 seats, so a small firm should expect the self-serve route. Whether the seats can be assigned to administrative staff, who may also benefit, is up to you, and a small firm often finds that giving every team member a seat makes the minimum less wasteful.

A worked decision for a six-person advisory firm

Take an advisory firm with three advisers, two paraplanners and one operations manager. They want the plugin for meeting prep, recaps and client-document checks. Three options stand out.

1. Team, six standard seats: $120 a month, $1,440 a year. No audit logs, no Compliance API, no custom retention. The firm writes its own rule that AI output is a draft, saves final records in the CRM, and keeps personal data to what the tasks need.
2. Enterprise, self-serve: at least $400 a month in seat fees, $4,800 a year, plus usage. Audit logs and the Compliance API become available. The firm assigns the spare 14 seats to colleagues who would otherwise not have access, and sets spend limits from the first day.
3. Wait on Claude: use a dedicated notetaker for meeting records and review the plugin at the next renewal, once the compliance officer has defined what AI records the firm must keep.

The decision rule I would use: if the firm's written compliance procedures, or its regulator's guidance as interpreted by your compliance officer, call for a record of AI use, choose Enterprise or option three. If they do not, Team can be defensible, but you carry the burden of explaining why the firm did not follow the vendor's own recommendation for regulated firms, so write the reasoning down. Neither answer is legal advice, and a compliance officer should sign off on whichever is chosen. The tutorial on what compliance allows when advisers use ChatGPT covers the equivalent question for a different assistant.

A first week with the plugin

1. Day 1, install and onboard: install the plugin from the plugin directory, then enter the prompt Anthropic gives: "Onboard me to Claude for Financial Advisors". The guided setup lets you choose which tools to connect. Anthropic says no IT department is needed.
2. Day 1, connect little: connect your Microsoft 365 account and your CRM first. Leave custodian and portfolio connections until the compliance officer has seen what they can read.
3. Days 2 and 3, test on past meetings: run the transcript-to-recap skill on five past meetings where the client has agreed to the use of AI, and compare each output with the notes you kept at the time.
4. Day 4, review: list every error: a wrong number, an invented commitment, a missed concern. Count them.
5. Day 5, decide: if errors are rare and easy to spot, plan a wider pilot on one adviser's book. If they are frequent or subtle, tighten the instructions before going further.

The Claude Team admin settings checklist and the tutorial on whether Claude Team is worth it help with the admin side of that week on the lower tier, and they apply in large part to Enterprise as well.

Two outputs and what a reviewer would fix

These are illustrative outputs from a pilot-style test, to show what review catches. The first is a CRM task list from a transcript.

Meeting recap (draft)
- Client agreed to increase pension contribution by $500 a month.
- Client will send the old workplace pension statement by Friday.
- Adviser to review protection cover at next meeting.

CRM tasks
1. Update contribution amount to $500 more (due: this week)
2. Chase statement (due: Friday)
3. Add protection review to next agenda

In the transcript, the client said they "would consider" the extra $500 once a bonus was confirmed. The recap turned that into an agreement and the CRM task would change a figure. The reviewer rewrites the line as "The client will consider increasing contributions once their bonus is confirmed", deletes task one, and adds "Ask about bonus confirmation". This is the main failure to guard against: a hedged remark that hardens into a commitment.

The second is a flagged draft email, where the compliance skill raises a concern.

Draft: "Our approach has consistently beaten the market, so we are
confident your portfolio will keep growing."

Skill flag: possible performance claim and implied guarantee.
Needs compliance review before sending.

The skill cannot decide, and in this case the reviewer removes the sentence and replaces it with the factual statement the firm uses in its approved text. The flag is useful because it prompts a second look. It does not replace the person whose job is to approve the wording, and Anthropic's own materials say it does not make compliance determinations. The tutorial on AI marketing for advisers that clears compliance goes through that review step in more depth.

Budget a one-month pilot before committing

A pilot costs much less than a year of the wrong plan. For the six-person firm above, a month on Team with monthly billing is six standard seats at $25, which is $150. The time budget is more important: about two hours for setup and the onboarding prompt, five meetings at roughly 30 minutes each to compare recaps with your notes (2.5 hours), and an hour at the end for the error count and the plan decision. That is under six hours of someone's time, plus $150.

The pilot answers three questions that a pricing page cannot. How many of the plugin's outputs does a reviewer have to correct? Which tasks does the firm actually repeat, so which skills are worth wiring to connectors? And does the compliance officer, having seen real outputs, want the record that only Enterprise provides? Anthropic's help centre describes an upgrade path that keeps the same organisation when a Team plan moves to Enterprise, preserving memberships, roles, conversations and projects plus some settings, so the pilot can be a stepping stone rather than a separate exercise. Before go-live, raise any spend limit set during the pilot: once the organisation-level limit is reached, everyone loses access until it is raised or the next month begins.

Keep the error count as a simple table in a spreadsheet: meeting, number of errors, type (number, commitment, omission, tone) and minutes to fix. A pattern in the type column tells you where to tighten the firm's instructions; a high minutes-to-fix total tells you the plugin is not saving time yet.

Questions for Anthropic and your compliance officer

• Does the firm's record-keeping duty extend to AI prompts and outputs, and which record satisfies it: audit-log events, chat content through the Compliance API, or the final note in the CRM?
• For each connector, what access level and licence does the vendor require, and does it charge for the connection?
• Where do the connector's data requests travel, and what does Anthropic retain? Ask for the data processing terms and the retention terms for your plan in writing.
• Is client consent needed before meeting transcripts are processed, and how is it recorded?
• Who reviews AI-drafted client communications, and where is that review evidenced?
• What happens to the data and the logs when a team member leaves, and how fast can an admin remove access?
• What spend limits will the firm set on Enterprise usage, and who watches them?
• How will the firm document its AI policy so it can show a regulator or auditor what the controls are?

Treat the answers as the firm's file on the decision. If one answer comes back unclear, that is a finding in itself, and it tells you which part of the pilot to slow down.

Five common misreadings of the launch coverage

• "The plugin is an AI adviser." It prepares drafts and briefs for the adviser, who remains responsible for advice and client communications.
• "You must have Enterprise to run it." Anthropic's page says it works on paid plans. Enterprise is the recommended plan for registered investment advisers because of audit logs.
• "Audit logs hold the conversation." They record events but not the titles or content of chats; the Compliance API is the route to content.
• "Connectors mean automatic trading or transfers." Anthropic says Claude does not place trades or move money.
• "The launch credit is still available." The one-time usage credit applied to Enterprise requests made by the end of September 2026, so it has lapsed. Check current terms for anything similar.

The heading on Anthropic's solutions page, "Focus on financial advice, not admin", is a fair summary of what the plugin is for. The plan question is how much control the firm needs around that delegation, and what it will pay for it.

Questions advisers ask about the Claude plugin and plans

Does the plugin only work on Claude Enterprise?

No. Anthropic's page says the plugin works on paid Claude plans. What it recommends for registered investment advisers is Enterprise, because that plan includes audit logs and compliance controls Team lacks. So Enterprise is the plan Anthropic suggests for a regulated firm, not a technical requirement to run the plugin.

Can a one-adviser firm buy Claude Enterprise?

Self-serve Enterprise needs at least 20 seats, billed annually at $20 a seat plus usage at API rates, so a one-adviser firm pays for 20 seats, at least $400 a month before usage. Sales-assisted plans need 50. A tiny firm usually starts on Team or waits until its compliance needs justify the minimum.

Does Claude give investment advice with this plugin?

Anthropic says it does not give investment advice, place trades or move money. The plugin prepares notes, emails, briefs and CRM updates that stay as drafts until the adviser approves them. Investment recommendations, client communications and compliance decisions remain with the human adviser and the firm.

Are Enterprise audit logs a full archive of AI conversations?

No. Anthropic's help page says audit logs record events such as sign-ins, file uploads and conversation creation over the past 180 days, but not the titles or content of chats. Chat histories and file content are available through the separate Compliance API, so ask your compliance officer which record the firm actually needs.

Further reads

• AI CRM for Financial Advisers: What Is Built In and What Is Bolted On — Which CRM AI features are built in and which are add-ons.
• Jump vs Zocks: Which AI Notetaker Fits Your Compliance Set-Up? — Meeting notetakers compared on compliance set-up.
• AI Financial Planning Software for Advisers: What to Test First — What to test first in AI planning software.
• How Advisers Use AI to Prepare for Annual Client Reviews — Preparing annual client reviews with AI.
• AI Automation for Financial Advisers: Onboarding Without Re-Keying — Onboarding clients without re-keying data.
• Wealthbox AI Notetaker Cost: Is It Cheaper Than Jump or Zocks? — Whether a CRM-native notetaker beats a standalone one.
• Zocks Pricing for Financial Advisers: What to Ask Before You Sign — What to ask a notetaker vendor about archiving.
• Best AI for Investment Management Firms: What to Pay For and Skip — A spend ladder for advisers: what to buy first, which note-taker and tax tools justify their price, and which AI purchases a small firm can skip.
• AI for Financial Advisers UK: Suitability Letters and Meeting Notes — A workflow for drafting suitability letters and meeting notes with AI while the adviser stays accountable and Consumer Duty evidence stays intact.
• AI for Asset Managers: Client Reports, Meeting Notes and Records — What a boutique asset manager can safely hand to AI, what stays human, and the record of AI-assisted work to keep for compliance.
• AI Tools and AI Development: The Complete 2026 Guide — the AI hub, including every tutorial in the AI-for-business series.

Sources: Anthropic's Claude for Financial Advisors solution page and launch post, the Claude pricing page, Claude help articles on the Enterprise plan, the Team plan and audit logs, and Anthropic's privacy centre page on data retention, read on 1 November 2026.