Claude for Financial Advisors: Preparation and Reviewed Drafts

Claude is an AI assistant made by Anthropic. Financial advisors can use it for background research, educational content drafts, meeting preparation questions, and communication drafts for compliance review. Claude does not have access to market data, client account systems, or regulatory databases, so every output that involves specific advice or client-facing content needs review before use.

Keeping approved procedures and reusable prompts in a shared record outside any one session makes this sustainable. Tools like fenbs let advisors record which prompts produce compliant drafts, which review steps apply to each type of output, and where each task stands. For tools that compare different AI assistants for advisory work, coverage of AI assistants for financial advisors addresses that question directly.

Research and background preparation with Claude

Claude can help advisors research a topic they need to understand before a client conversation: how a type of investment structure works, what a regulatory change requires, or what questions advisors typically ask when reviewing a client's estate planning documents. It draws on its training data, not live sources, so it is most reliable for well-established concepts and principles.

For current regulations and recent guidance, verify against primary sources: the SEC, FINRA, IRS, or the relevant state regulator. Claude's knowledge has a training cutoff date, and financial regulations change. Treat Claude's explanation of a rule as a starting point for your own research, not as a substitute for checking the current text.

Drafting educational content

Advisors who produce educational newsletters, blog posts, or explainer documents for clients can use Claude to draft those materials. Claude can explain concepts such as sequence-of-returns risk, tax-loss harvesting, or Roth conversion strategies in plain language for a general audience.

Educational content for clients is still subject to FINRA and SEC communication rules. A broker-dealer principal must review communications before they are distributed. The review confirms that the content is balanced, accurate, and not misleading. Claude's draft reduces the time to a reviewable draft; the compliance review remains required.

Advisors at independent RIAs operate under the SEC's communication rules rather than FINRA's. The review requirement is similar in principle: the content must be accurate, not misleading, and consistent with the firm's policies. Check with your compliance officer for the specific review process that applies to your firm.

Meeting preparation

Claude is useful for preparing for a client meeting without using any client information. Ask it to generate a list of questions an advisor typically asks when reviewing a client's retirement readiness, or to explain the key considerations in a particular product category so you can discuss them confidently.

You can also use Claude to prepare for difficult conversations: ask it to suggest how to explain a portfolio drawdown in plain terms, or to list the common concerns clients raise about a specific type of allocation decision. That preparation helps advisors run more effective meetings without any client data entering the AI tool.

Compliance review before any client use

Nothing Claude drafts should be sent to a client or used in a client recommendation without compliance review. Claude is an AI assistant, not a compliance-approved tool. It does not know your firm's supervisory procedures, your client's suitability profile, or the specific communication standards your firm has established.

The practical rule is: Claude assists the preparation process, a qualified person reviews the output, and the advisor takes responsibility for anything that goes to a client. That is the same standard that applies to any draft an assistant produces, whether the assistant is human or AI.

Building reusable prompts for advisory work

The same types of questions come up repeatedly in advisory practice: questions about tax treatment, estate planning concepts, product comparisons, and educational explanations of common planning strategies. Writing a good prompt once and reusing it is more efficient than rewriting from scratch each time.

A good reusable prompt for an educational piece includes the topic, the audience (a client who is not a financial professional), the length, the tone, and any specific points to include or avoid. Saving that prompt with the type of content it is designed for means the next time you need a similar draft, the starting point is already there.

What Claude does well for advisors, and what it does not

Claude is well suited to explaining concepts, drafting educational content, generating meeting preparation questions, and writing a first version of a communication that a person will review. These tasks are low-risk starting points because the advisor reviews the output before it reaches a client.

Claude is not suited to receiving non-public client information: account values, holdings, tax details, or personal financial data. Claude is a general AI assistant and its data handling is governed by Anthropic's policies, not by your firm's information security or privacy requirements. Use client information only in systems approved for that purpose by your firm.

Claude also should not produce specific investment recommendations or advice for a named client. It does not know the client's situation, risk tolerance, time horizon, or suitability profile. That determination is the advisor's responsibility, supported by the firm's suitability analysis process, not by an AI assistant.

Comparing AI assistants for advisory work

Claude is one of several AI assistants advisors use. Others include ChatGPT, Gemini, and Microsoft Copilot. For most advisory preparation tasks, the differences in capability are less important than the differences in data handling, enterprise agreements, and your firm's approved tools list.

If your firm has an enterprise agreement with one AI provider, use that tool for work that involves your firm's systems. If you are using a personal AI account for preparation work, ensure that no client information enters the tool. The choice of model matters less than the data handling question.

Coverage of what Claude is and how its API works are useful starting points if you want to understand what Claude can and cannot do technically before deciding how to use it in your practice.

Building a sustainable AI-assisted advisory practice

The advisors who get the most from Claude are the ones who build a consistent process rather than using it ad hoc. Ad-hoc use means rewriting the same prompts from memory, rediscovering the same limitations, and starting each new task without the benefit of what worked before. A consistent process means starting from tested prompts, applying the same review steps, and recording what was approved. That record of approvals is also the starting point for the next review: if the same type of communication was reviewed and cleared last month, the next draft of that type starts from a higher baseline.

Start with one use case: either educational content drafts or meeting preparation, whichever is the bigger time drain in your practice. Write the prompts that produce the output you want, run them through the compliance review process once, and note what the reviewer adjusted. Those adjustments tell you what the prompt was missing. After five or six rounds of improvement, the prompt produces drafts that need less editing, and the review step becomes predictable in how long it takes.

Then extend to a second use case, applying the same approach. Build one reliable process before adding more. An advisory practice with two well-developed AI workflows, each with tested prompts, a documented review step, and a record of what was approved, is more productive than one with eight half-finished uses that each require significant manual correction before the output is usable.

How to do this with fenbs

Set up a project for your advisory practice. Record the approved review procedures for each type of Claude-assisted output, save the prompts that work well, and track tasks without including client information.

1. Create a free account. A personal board works for a solo practice; a team board lets support staff and their AI assistants see the same process decisions and tasks.
2. Add a project for your practice. Under Decisions and rules, record the review steps that apply to each type of output: which prompts require principal review, which are for internal use only.
3. Save your best-performing prompts under Skills and prompts. Note the intended use and any constraints for each one.
4. Add tasks for content or preparation work in progress. Keep descriptions general enough that they do not include client names or account details.
5. Before ending a session, save progress so the next session picks up from the current state of the work.