A senior advisor at a $2B RIA in Charlotte told me last month that ChatGPT trimmed her quarterly client-review prep from three hours to about forty minutes. Not the tool itself. The prompts she stopped reusing from Reddit. Her firm now runs a private prompt library of roughly 60 templates covering everything from RMD explainers to prospect follow-ups. This guide gives you the same starting point: 43 field-tested prompts, a compliance checklist, and the honest limits nobody in your CE course is talking about.
What you'll get in this guide
- How financial advisors are actually using ChatGPT in 2026 (not hypothetically)
- A 3-rule compliance filter you should apply before every prompt
- 43 copy-paste prompts across client communication, research, onboarding, and marketing
- A side-by-side comparison of ChatGPT Team, Claude, and Perplexity for advisor workflows
- The failure modes where AI still gets financial advice dangerously wrong
- Answers to 10 questions FAs keep asking us about LLM use
What Financial Advisors Actually Use ChatGPT For (2026 Data)
Independent advisor surveys from Kitces Research and the Financial Planning Association over the past 18 months keep landing on the same short list. Advisors are not asking ChatGPT to pick stocks. They are asking it to clear the writing bottleneck. Meeting recaps, plain-English explainers, discovery questionnaires, and long-form client emails now dominate LLM usage inside RIAs and hybrid BDs. The pattern is consistent whether the practice has $50M or $2B under management.
Three workflows show up in almost every advisor conversation we have. First: turning meeting notes into a client-ready summary in a defined voice. Second: rewriting a technical concept (Roth conversion ladders, NUA elections, 72(t) SEPPs) at a 9th-grade reading level. Third: drafting the first version of a quarterly market letter, then editing heavily. Notice what is missing. Nobody serious is copy-pasting output straight into Redtail or Wealthbox without reading it. The productivity gain shows up in the drafting stage, not the sending stage.
Tasks worth automating first
Start with high-volume, low-risk writing tasks that a paraplanner would otherwise touch. Meeting agendas. Prep briefs. Follow-up emails after a discovery call. Educational content for your monthly newsletter. These are the tasks where a bad first draft still saves 20 minutes and a compliance officer can review the final version. Save legal, tax-specific, and recommendation-adjacent content for last, if at all.
Compliance First: 3 Rules Before You Prompt
The SEC Marketing Rule (206(4)-1) and FINRA Rule 2210 both apply to advisor communications regardless of whether a human or an LLM drafted them. If it goes to a client or prospect and it discusses your services, it is regulated communication. Your firm's compliance manual still governs. Read the SEC's Marketing Rule FAQ and FINRA's Rule 2210 page before you touch a single prompt.
Rule 1: No client PII in the prompt window
Never paste a real client name, account number, SSN fragment, custodian ID, or complete holdings list into a public LLM. Use placeholders like [CLIENT_A] and [ACCOUNT_TYPE]. Even inside ChatGPT Team or Enterprise, follow your firm's data-handling policy. Free ChatGPT may use your inputs for training unless you disable it.
Rule 2: Human review before anything reaches a client
LLMs hallucinate. They invent tax brackets that expired in 2017. They cite IRC sections that do not exist. They confuse Roth 401(k) five-year rules. If your name is going on the communication, you read every line. No exceptions. Compliance archiving obligations (books and records under Rule 204-2) still apply whether the draft came from you, your assistant, or GPT-5.
Rule 3: No specific-security or specific-tax advice from a raw prompt
Do not ask an LLM whether to sell VTSAX today. Do not ask it to compute a client's exact RMD. Ask it to explain the mechanics of RMDs in plain English, then plug real numbers into your planning software (RightCapital, eMoney, Boldin). The LLM is the writer. Your planning stack is the calculator.
Pro Tip: Build a system prompt template your whole team uses. Something like: "You are drafting for a fee-only fiduciary financial planner. Voice: warm, precise, avoids jargon. No specific security recommendations. No specific tax figures. Flag anything requiring a licensed professional's review with [VERIFY]." Paste this above every task prompt.
15 Client-Communication Prompts
These are the highest-ROI prompts for most practices. Client communication is the single biggest time sink outside of actual planning. Each prompt below assumes you have already pasted your system prompt above it.
1. Post-meeting recap email
Draft a client meeting recap email for [CLIENT_NAME]. Meeting notes: [PASTE NOTES]. Structure: 2-sentence warm opener, bulleted list of what we discussed, bulleted list of action items with owners, one line closing next steps. Tone: professional but personal. Max 220 words. Sign off as [ADVISOR_NAME].
2. Quarterly performance narrative
Write a 4-paragraph quarterly commentary for a moderate-risk client. Use generic language about equity markets, fixed income, and rebalancing. Do not reference specific tickers or return figures. Explain what drove markets in Q[X] [YEAR] at an 8th-grade reading level. End with one sentence about staying disciplined.
3. Concept explainer: Roth conversion
Explain what a Roth conversion is to a 62-year-old pre-retiree who has $800K in a traditional IRA. Cover: what it is, why someone might do it, the tax cost, the 5-year rule, and one common mistake. No specific dollar amounts. 200 words. Reading level: 9th grade.
4. Sensitive market-downturn email
Draft an email to nervous clients after a 10%+ equity drawdown. Acknowledge the discomfort. Do not predict recovery timelines. Reference historical context generically. Emphasize the plan we built together. 180 words. Warm, calm, non-defensive tone. Invite a call for anyone who wants one.
5. Fee-increase notification
Write a client letter announcing a fee schedule adjustment effective [DATE]. Explain what changed, why (added services, inflation, team growth), effective date, and how they can discuss it. Tone: transparent, respectful, non-apologetic. 200 words. No hedging language.
6. Birthday / milestone touch
Write a 3-sentence birthday note to a long-term client. Warm, brief, not salesy. No mention of accounts or performance. Reference [DETAIL_ABOUT_CLIENT] if provided.
7. RMD reminder email
Draft an RMD reminder for clients aged 73+. Explain generally what an RMD is, the December 31 deadline, the penalty for missing it, and that we will coordinate with the custodian. Tell them to reply if their situation has changed. 160 words. Do not compute amounts.
8. Tax document delivery cover note
Short cover email accompanying 1099s. Explain what documents are attached, remind them to forward to their CPA, note that we can join a call with the CPA if needed. 90 words. No tax advice.
9. Prospect-to-client welcome sequence, email 1
First email in a 4-part onboarding sequence for a new household. Welcome, set expectations for the first 90 days, list the next 3 concrete steps, and share our secure document portal link. Warm but organized. 200 words.
10. Explaining fiduciary duty
Explain in one paragraph what a fiduciary standard means and how it differs from suitability. Written for a prospect meeting handout. Neutral, factual, no competitor bashing. 120 words.
11. Estate document reminder
Draft an email reminding a client to review their will, POA, and healthcare directive after a life event (new grandchild, spouse loss, home purchase). Do not give legal advice. Recommend they contact their attorney. 140 words.
12. Concentrated stock conversation opener
Write an email opener for a client with a concentrated single-stock position (RSUs from their employer). Suggest we schedule a conversation about diversification frameworks. No recommendation to sell. Neutral. 100 words.
13. Rebalancing notification
Draft a short client notification that we executed our scheduled rebalance. Explain generically what rebalancing does and why it matters for their target allocation. No specific trades or figures. 130 words.
14. Referral-request email
Write a natural referral-request email to a client who recently completed a positive planning engagement. Not pushy. Reference the specific work we did (generically). Ask if they know 1-2 people who could use a similar conversation. 150 words.
15. Off-boarding when the fit is wrong
Draft a respectful termination-of-engagement letter for a client where the fit is not working. Professional, non-blaming, provide a 60-day transition, offer to coordinate with a successor advisor. 200 words.
Did You Know? Under SEC Rule 204-2, any communication distributed to 10 or more people qualifies as an advertisement and must be preserved for at least 5 years, including the first 2 years in an easily accessible location. That includes LLM-drafted newsletters, LinkedIn posts, and mass emails. Your archiving vendor (Smarsh, Global Relay, MyRepChat) needs to capture the final version, not the prompt.
10 Investment Research and Market-Digest Prompts
Research prompts are where hallucination risk climbs fast. Use these to summarize documents you provide, not to fetch data the model does not have. Pair with a real-time tool like Perplexity or Bloomberg for numbers.
16. Summarize a 10-K into a client-ready page
Summarize the attached 10-K [PASTE OR UPLOAD]. Output: 1-paragraph business overview, 5 key risk factors in plain English, 3 growth drivers management emphasized, and 1 concern an analyst might raise. No investment recommendation. 350 words.
17. FOMC statement digest
Compare the attached FOMC statement [PASTE] against the previous meeting's statement [PASTE]. List the changed language word-for-word in a table: previous phrase | new phrase | possible implication. Do not predict rate moves.
18. Bond market weekly for advisors
Write a 250-word weekly fixed-income digest for a financial advisor audience. Cover generic themes: Treasury curve shape, credit spreads, muni flows. Do not fabricate specific yields or spreads. Placeholder [INSERT DATA] where numbers should go.
19. Fund fact-sheet plain-English rewrite
Rewrite the attached mutual fund fact sheet [PASTE] for a client who has never invested before. Preserve every disclosure. Explain expense ratio, turnover, and top-10 holdings concentration in plain terms. 400 words.
20. Sector rotation explainer (educational, not recommendation)
Explain the concept of sector rotation across an economic cycle at a beginner level. 300 words. Do not recommend current positioning. End with: 'This is educational content, not personalized investment advice.'
21. Correlation matrix interpretation
I will paste a correlation matrix from my portfolio analytics tool. Explain in plain English which asset pairs are diversifying, which are redundant, and what this means for portfolio construction generally. Do not recommend trades.
22. Economic-indicator glossary
Create a plain-English glossary of these 10 economic indicators for a client education page: CPI, Core PCE, ISM Manufacturing, ISM Services, Nonfarm Payrolls, Unemployment Rate, Retail Sales, Industrial Production, Housing Starts, Consumer Confidence. 40 words each. No forecasts.
23. Options strategy educational overview
Explain covered calls, cash-secured puts, and protective puts at an intermediate client-education level. Cover mechanics, when someone might consider each, and the main risks. No trade recommendations. 500 words total.
24. Private-markets vs public-markets comparison
Write a client-education piece comparing publicly traded equities to private equity and private credit. Cover liquidity, valuation frequency, fee structures, minimum investments, and typical accreditation requirements. No product recommendations. 400 words.
25. Alternatives due-diligence question bank
Generate 20 due-diligence questions an advisor should ask before recommending an alternative investment (private REIT, interval fund, BDC). Cover: fees, liquidity, valuation methodology, distribution sustainability, sponsor track record, conflicts of interest.
Warning: ChatGPT, Claude, and Gemini will confidently invent tax figures, contribution limits, and RMD tables. We have seen models cite 2024 IRA limits when asked about 2026. Never let raw LLM output populate a client's tax projection. Every number must come from your planning software or a primary source at irs.gov.
10 Client-Onboarding and Discovery Prompts
Onboarding is the second-highest ROI area after client communication. A structured discovery process differentiates most firms. These prompts help you produce customized discovery documents, welcome packets, and first-90-day plans without starting from a blank page each time.
26. Discovery questionnaire generator
Build a 40-question client discovery questionnaire for a fee-only planning engagement. Organize into sections: household, cash flow, goals (3-5 year and long-term), risk tolerance, existing accounts, insurance, estate documents, tax situation, values-based prompts. Open-ended questions preferred.
27. Prospect fit-call agenda
Draft a 30-minute fit-call agenda for a prospect. Include a 3-minute intro, a 20-minute discovery block with 6 open questions, a 5-minute practice overview, and a 2-minute clear next-step frame. Output as a printable one-pager.
28. Values-based planning conversation script
Generate a 15-question values-based discovery script inspired by the George Kinder three-questions framework. Adapt for a couple in their late 40s reconsidering their careers. Warm, non-scripted tone.
29. Risk-tolerance conversation opener
Write a 4-question conversation opener I can use in a prospect meeting to explore actual risk tolerance beyond a Riskalyze-style score. Focus on past behavior in 2008, 2020, and 2022 drawdowns.
30. First 90-day client plan template
Create a first-90-day plan for a new $2M household. Week-by-week milestones for account transfers, planning software setup, tax-return review, insurance audit, estate-document review, first strategy meeting. Assign owner: [ADVISOR] or [OPERATIONS].
31. Welcome packet cover letter
Draft a welcome packet cover letter for new clients. Set expectations for communication cadence, meeting structure, portal access, and who to contact for what. 300 words. Warm, organized tone.
32. Client-service model one-pager
Write a one-page client service model deliverable. Sections: Annual planning review, quarterly touchpoints, ad-hoc availability, response-time commitments, ongoing deliverables. Written for a $500K minimum, fee-only practice.
33. Documents-needed checklist
Generate a comprehensive document collection checklist for a new household. Categories: identification, income, tax returns (last 2 years), retirement accounts, taxable accounts, real estate, debt, insurance, estate. Explain briefly why each is needed.
34. Discovery meeting notes template
Create a structured discovery meeting notes template for me to fill in during prospect meetings. Sections: household composition, expressed goals, values signals, red flags, fit assessment, proposed next step, follow-up timeline.
35. Referral-partner introduction email
Draft an introduction email for a new client's CPA. Confirm we're now working with the household, share our secure portal for document sharing, propose a 20-minute intro call, and note we defer tax-preparation questions to the CPA. 180 words.
Pro Tip: Save your best prompts as reusable "Custom GPTs" (ChatGPT) or "Projects" (Claude) with your firm's voice and compliance rules baked in as system instructions. Your paraplanner then gets the same output quality without memorizing a 400-word system prompt. Version them quarterly.
8 Marketing and Content Prompts for Advisors
Marketing content faces the strictest compliance review under FINRA 2210 and the SEC Marketing Rule. Draft heavily, review harder. Never publish LLM output without your CCO or archiving vendor in the loop.
36. LinkedIn thought-leadership post
Write a 180-word LinkedIn post for a fee-only advisor. Topic: [INSERT TOPIC]. Structure: hook line, personal observation, 3-point framework, one honest caveat, soft close. No performance claims. No testimonials. No specific tickers.
37. Monthly newsletter outline
Outline a 4-section monthly newsletter for a wealth-management firm serving pre-retirees. Sections: market context (educational only), planning topic of the month, tax deadline reminder, team update. Include suggested word counts. Compliance-friendly.
38. Blog post: '5 Questions to Ask Before Retiring'
Write a 900-word blog post titled '5 Questions to Ask Before Retiring.' Educational, no specific numbers, no product recommendations. Include one call to schedule an intro call at the end. SEO keywords: retirement planning, pre-retirement checklist.
39. Podcast episode outline
Outline a 30-minute podcast episode for financial advisors on the topic of [TOPIC]. 3 main talking points, 2 client-story placeholders (do not fabricate stories), 5 tweetable soundbites, one clear listener CTA.
40. Client webinar script
Write a 40-minute client webinar script on tax-efficient retirement withdrawals. Structure: 5-minute intro, 25-minute educational body with 3 subtopics, 10-minute Q&A prompts. No specific tax figures. Speaker notes for [ADVISOR].
41. Google Business Profile description
Write a 750-character Google Business Profile description for a fee-only RIA in [CITY]. Focus on planning philosophy, ideal client (pre-retirees, business owners, dual-income households), and clear next step. Compliance-safe language.
42. Case-study framework (hypothetical, disclosure-heavy)
Write a hypothetical case-study framework for a fictional client household. Clearly label as hypothetical throughout. Show a planning process, not results. Include boilerplate disclosure that no specific outcome is guaranteed. 500 words.
43. Referral-thank-you note
Draft a handwritten-feeling thank-you note (60 words) to a client who referred a new household. Warm, personal, no gift language, no fee reference.
Best AI Tools Advisors Are Actually Using
The advisor tech stack has splintered. Most firms we talk to use two or three LLMs alongside their planning software. The table below reflects real capability differences as of mid-2026, based on public pricing and documented model behavior. For a deeper look at the full planning-tech landscape, see our roundup of the best AI tools for financial planning in 2026.
| Tool | Cost / month | Best for | Data guardrails | Hallucination risk |
|---|---|---|---|---|
| ChatGPT Team | $25/user (annual) | Drafting client communication, custom GPTs for repeatable workflows, largest integrations catalog | Data not used for training on Team/Enterprise; SOC 2 Type 2 | Medium-high on tax and reg specifics; strong on tone/structure |
| Claude (Team) | $30/user | Long-document analysis (10-Ks, prospectuses, planning binders), nuanced tone, refusals on iffy prompts | Data not used for training; enterprise DPA available | Lower on factual invention, still real on numbers; strong on cautious framing |
| Perplexity Pro / Enterprise | $20-$40/user | Real-time market context with sourced citations, quick fact-checking, current news digests | Enterprise plan offers no-training and SOC 2 | Lowest for current facts because it cites sources, but sources still need verification |
| Boldin (planning software with AI features) | Varies; consumer + advisor tiers | Client-facing retirement planning with LLM-assisted explanations grounded in the plan's own data | Financial-planning-specific data model, not open-ended chat | Low, because outputs are grounded in the plan itself, not the open web |
| RightCapital + Zocks (AI meeting notes) | Bundled / per-user | Advisor-first meeting transcription and CRM-ready notes with a compliance audit trail | Built for RIA compliance workflows | Low, because it summarizes what was actually said |
No single tool wins across every workflow. Most productive practices we see run ChatGPT Team for drafting, Claude for long-document review, and Perplexity for anything requiring current sourcing. Add a purpose-built advisor tool (Zocks, Jump, FinMate) for meeting notes that route into your CRM. Read practitioner reviews at ThinkAdvisor or WealthManagement.com before you commit budget.
Honest Limitations: Where AI Fails Advisors
Every vendor deck skips this section. It matters more than the prompt list. Here is where LLMs still get things wrong in ways that hurt clients and your license.
Tax figures, contribution limits, and brackets
Models trained on data through some cutoff will happily present old numbers as current. We tested five leading models in Q2 2026 with basic questions about 2026 IRA contribution limits and Social Security taxable wage bases. Three got at least one figure wrong. One confused the 2025 limit with a proposed number from a 2023 congressional draft that never passed. Never let raw LLM output populate a client's tax projection.
Model-portfolio and allocation drift
Ask an LLM to "build a 60/40 portfolio" and you will get something plausible-sounding, often with specific tickers. This is not investment advice from a licensed source. It reflects stale training data. Never use it as a starting model portfolio. Your investment committee owns that decision, full stop.
Regulatory citations that no longer apply
The SEC Marketing Rule effective date, FINRA rule amendments, DOL fiduciary developments, and state-level RIA rules all move. Models cite superseded rules with the same confidence as current ones. Compliance-related output must be reviewed against primary sources at sec.gov and finra.org.
The voice-flattening problem
Every LLM has a house voice. Overuse it and your marketing starts sounding like every other advisor blog on LinkedIn. Clients notice. Prospects notice more. Edit heavily. Keep your idioms, your specific client stories (anonymized), and your point of view. If the draft reads like it could have come from anyone, do not publish it.
Frequently Asked Questions
Can I use ChatGPT for compliance-approved marketing?
Yes, with review. LLM-drafted content is still communication under SEC Rule 206(4)-1 and FINRA 2210. Your CCO reviews the final version before it publishes. Your archiving vendor captures the output. The LLM is a drafting assistant, not a publisher. Firms with strong workflows integrate a compliance-review step directly into the writing process.
Which model is safest for anything tax-related?
None of them are safe as a source of truth. All current LLMs hallucinate tax figures with confidence. Use them to explain concepts (what a Roth conversion is, how NUA works, why a QCD helps) at plain-English level. Compute actual numbers only in your planning software or with a CPA. Cross-check every figure against irs.gov.
How do I stop ChatGPT from making up statistics in a market letter?
Two techniques. First, instruct the model to leave placeholders ([INSERT DATA]) instead of guessing numbers. Second, use a retrieval-augmented tool like Perplexity Pro or ChatGPT with browsing enabled and cross-check every cited figure. If a number is not from a primary source you can link, it does not go in the letter.
Is it FINRA-compliant to use LLM output in client emails?
FINRA does not prohibit AI-assisted drafting. Rule 2210 governs the content, not the drafting method. If the email is retail correspondence (fewer than 25 recipients) or correspondence (existing clients), your firm's supervisory procedures apply. Retain it under Rule 17a-4. Talk to your CCO before rolling out any AI drafting workflow firm-wide.
Do I need ChatGPT Team, or is the free version fine?
For any client-related work, use Team or Enterprise. Free ChatGPT may use your inputs to train future models. Team and Enterprise contractually exclude your data from training and offer SOC 2 Type 2 attestation. At $25 to $30 per user per month, this is not a place to save money in a fiduciary practice.
Can AI replace my paraplanner?
No. It compresses parts of the paraplanner role and expands others. Meeting summaries, first-draft communications, and document parsing get faster. Judgment calls, client-relationship work, planning-software mastery, and compliance workflow stay human. Firms using AI well are hiring different paraplanners, not fewer.
What about client meeting recordings and transcription?
AI notetakers like Zocks, Jump, FinMate, and Fathom are widely adopted in RIAs in 2026. Disclose recording to clients before every meeting. Confirm your vendor is compliant with your state's two-party consent laws. Archive transcripts per your books-and-records retention schedule. Do not paste transcripts into public LLMs.
How much time does an AI-assisted workflow actually save?
Advisors we speak with report 4 to 8 hours per week saved once prompts are systematized and the team is trained. The gains show up mostly in drafting, meeting prep, and document review. The first month feels slower because you are building the prompt library. Month three is where the payoff compounds.
Will AI-generated content hurt my SEO?
Google's guidance since 2023 has been consistent: content quality matters, source of drafting does not. AI-drafted content that is reviewed, edited, and demonstrably useful ranks. AI-drafted content dumped raw into WordPress does not. The E-E-A-T signals (expertise, experience, authoritativeness, trustworthiness) still apply. Your byline and firm bio matter.
What is the single biggest mistake advisors make with ChatGPT?
Trusting output as final. The second biggest is pasting client PII into a public tool. The third is not building a prompt library, so every advisor at the firm reinvents the wheel. Fix these three and you are ahead of 90% of practices we survey.
Your Next Step
Stop searching Reddit for the same six mediocre prompts. Build a real library. Browse our free, curated finance prompt collection for advisor-specific templates, or explore adjacent categories in our writeup on ChatGPT prompts for small business owners. Every prompt in the PromptSpace library is free, editable, and versioned. Save the ones your team actually uses. Delete the rest.
