Between January 2024 and May 2026, AI note-takers ate the financial-advisor calendar. The Schwab 2026 RIA Benchmarking Study reports that AI adoption inside the RIA channel more than doubled versus 2023, and the category breakdown is unambiguous — meeting AI is the leading entry point. Jump and Zocks dominate the Kitces / Nerd's Eye View March 2026 AdvisorTech map; FinMate AI, Sybill, and Zeplyn round out the named field; RFG Advisory's enterprise investment in Zocks signaled the BD / aggregator buy-in; the Morgan Stanley and Merrill wirehouse pilots have moved from quiet trials into named home-office programs. The productivity claim — "10+ hours per week saved" — is the Zocks-published figure that survived two years of advisor scrutiny and is now corroborated across the category. Eighteen months ago, an advisor explaining AI note-taking to a colleague had to define the category. By May 2026, an advisor who hasn't deployed meeting AI is the outlier. This lesson is the practitioner-grade map of what meeting AI does, what it misses, how the workflow actually reshapes the advisor day under FINRA Rule 4511 retention, and how to read a vendor demo without falling for category-blur.

The Eighteen-Month Category Takeover

The pre-AI advisor day for a 200-household practice ran on a deeply unsatisfying ratio: every client meeting hour produced roughly 90 minutes to two hours of downstream administrative work — pulling the notes the advisor scribbled (or didn't), drafting the follow-up email, logging the activity in Wealthbox or Redtail, queuing the trade authorization in the planning software, updating the IPS section if anything changed, archiving the recording (if there was one) into Smarsh under FINRA Rule 4511, and scheduling whatever the meeting produced — a tax-return review, an estate-attorney introduction, a beneficiary update, a Roth conversion queue. The math compounded: six meetings in a day produced nine to twelve hours of downstream work, on top of the six meeting hours themselves, on top of the prospect calls, the prep, the compliance review, the CCO check-ins, and the quarterly commentary. Kitces' time-and-task research documented the canonical bucket split year after year — roughly 36% of an advisor's working hours sit in meeting prep, planning, and servicing — and the implication compounded into the productivity crisis the industry has been writing about since 2022.

Then meeting AI arrived. Jump and Zocks launched into a category that didn't exist as a category — there was no "AI note-taker for advisors" tag in the Kitces AdvisorTech map until late 2023. By March 2026, the map's meeting-AI cluster is the most-trafficked tile. The Schwab 2026 study's adoption data point — RIA AI adoption more than doubled vs. 2023 — is anchored heavily in this category; advisors are not running RAG pipelines on their first AI deployment, they are installing a Zoom / Google Meet / Teams overlay that captures the call, transcribes it, summarizes it, extracts action items, and drops the package into the CRM. The deployment friction is low, the productivity reveal is large, and the gateway effect into planning AI, tax AI, and CRM AI is now well documented.

The RFG Advisory enterprise investment in Zocks during 2025 marked the inflection where category buyers were no longer solo RIAs but networks, OSJs, and aggregators. The Morgan Stanley and Merrill wirehouse pilots — moving from "we're testing this" in 2024 to named home-office programs in 2025-2026 — moved the wirehouse channel from skeptical observation to active deployment. The Zocks-published "10+ hours per week saved" figure is now corroborated across vendor case studies and independent advisor surveys; the Financial Planning headline "Jump and Zocks Top Category as AI Notetaker Adoption Soars" captures the May 2026 landscape with no exaggeration.

The Named Five — Jump, Zocks, FinMate AI, Sybill, Zeplyn

Vendor-by-vendor positioning matters because the five named products are not interchangeable, and reading a vendor demo without the positioning map produces buyer's-remorse stories that pollute the category. The Kitces AdvisorTech map's March 2026 update is the canonical visualization; the positioning notes below are the practitioner-grade reading of where each fits.

Jump

Jump's positioning is "AI assistant for financial advisors" — meeting capture is the entry point, but the platform extends into pre-meeting prep brief generation, post-meeting follow-up email drafting, CRM activity logging into Wealthbox / Redtail / Salesforce FSC, and structured-data extraction back into the planning software (RightCapital, eMoney, MoneyGuidePro). Jump's adoption profile in 2026 is heavily independent RIA and ensemble firm, with growing OSJ adoption. The product's strength is the end-to-end workflow integration; the watch-out for the buyer is the standard category trap that every "AI assistant" platform produces — feature breadth that requires advisor discipline to consume without sprawling the prompt library.

Zocks

Zocks' positioning is more squarely "AI for advisor meetings" with deep emphasis on accuracy and compliance integration. The RFG Advisory enterprise investment in 2025 anchored Zocks as the category leader for the BD / aggregator / network buyer; the product's archive integration with Smarsh and Global Relay under FINRA Rule 4511 is the most-cited operational reason mid-sized BDs select Zocks over alternatives. The "10+ hours per week saved" figure originated in Zocks' published case-study data and has been the most-quoted productivity benchmark in the category. Zocks' adoption profile in 2026 spans independent RIAs, hybrid BD/RIAs, and increasingly the wirehouse channel via the named pilots.

FinMate AI

FinMate AI's positioning is "AI note-taker built for the advisor's exact meeting cadence" — discovery, annual review, prospect, and quarterly call templates with field structures pre-mapped to the workflows. The product appeals to firms wanting deeper template control without building it themselves and is frequently chosen by ensemble RIAs that want the field structures to enforce internal house-style consistency across multiple advisors. Integration depth with Wealthbox and Redtail is the most-cited operational anchor.

Sybill

Sybill originated in the sales-call analytics space and extended into wealth; the product's strength is sentiment, tone, and engagement analytics layered on the transcript — useful in the prospect and discovery context for advisors paying attention to the buyer-readiness signal. The watch-out is the same as for any sentiment-AI product crossing from sales into wealth: the Marketing Rule 206(4)-1 implication of using a buyer-readiness score to qualify or de-qualify prospects must be considered, and the firm's WSPs need to absorb it.

Zeplyn

Zeplyn's positioning is "AI for advisor productivity" with strong emphasis on the post-meeting follow-up workflow — the email draft, the action-item routing, the CRM update, and the calendar request for the next meeting. Zeplyn's adoption profile in 2026 favors firms whose pain point is the post-meeting SLA collapse (the six-meeting-day where four meetings have undocumented follow-ups by 5 pm). The product's structured-output integration into Wealthbox and Redtail is the operational anchor.

Adjacent — Pulse360

Pulse360 sits adjacent to the meeting-AI category, specializing in meeting prep and follow-up. It integrates with the major CRMs and is named in the AdvisorTech map; advisors evaluating meeting AI commonly evaluate Pulse360 alongside the named five, even though Pulse360's primary product surface is the prep-and-follow-up workflow rather than the live capture.

What Good Meeting AI Captures

The category capability that justifies the productivity claim is real, and the workflow reshape is the headline. A well-configured meeting AI in a 2026 advisor practice captures, at minimum, six artifacts from a single meeting hour: the verbatim transcript with speaker diarization (which speaker said what), a structured summary organized by topic and time-stamped, the extracted action-item list with owner and SLA, the next-meeting agenda items the advisor and client agreed to, the compliance flags (any client statement that suggests suitability concern, any potential conflict, any prospective change to the IPS, any KYC update), and the export package for the downstream systems — CRM activity log, planning software task, custodian form pre-fill, and the Smarsh / Global Relay archive entry under FINRA Rule 4511.

The end-to-end follow-up workflow that good meeting AI enables looks like this: meeting ends at 10:25. The transcript and summary are ready by 10:27. The advisor opens the platform on the way to the next meeting, scans the summary (90 seconds), accepts the action-item list with one minor edit, and the post-meeting routing executes — Wealthbox activity logged, follow-up email drafted to client (advisor reviews and sends), trade authorization DocuSign queued, planning software task created, Smarsh archive entry confirmed. Total advisor time post-meeting: roughly four minutes. The 90-minute-to-two-hour downstream cost of the pre-AI era has collapsed by roughly an order of magnitude. The Zocks-published "10+ hours per week saved" figure is exactly this collapse, summed across six meeting days. The Schwab 2026 data on adoption doubling is what happens when the collapse becomes visible to the next-cohort advisor.

What Meeting AI Misses — The Five Failure Modes

Meeting AI is a category-changing productivity lever. It also misses things, and the things it misses cluster in predictable failure modes the advisor must learn to compensate for explicitly. The mature firm bakes these compensations into the post-meeting workflow.

Whiteboard Math and Visual Reasoning

The most-cited missed artifact is whiteboard math — the moment in a discovery meeting where the advisor sketches the Social Security claiming decision on the whiteboard with two age columns and a breakeven analysis, or the Roth conversion staircase the advisor draws to show the bracket-fill logic visually. The AI captures the audio of "and you can see here that delaying to 70 makes sense for the higher earner because of the 8% delayed retirement credit..." but does not capture the drawing the advisor is pointing to. The compensation is a photo of the whiteboard (or a screenshot of a shared screen) attached to the meeting record, and a brief verbal description spoken explicitly for the transcript ("for the record, I'm showing the Hendersons a breakeven chart between age 67 FRA claiming and age 70 delayed claiming, with the breakeven occurring at approximately age 82.5").

Off-Mic Spouse and Side-Conversation Comments

The second failure is the off-mic comment from the spouse who isn't on camera — the "she's worried about Mom's care costs, but she doesn't want to bring it up yet" comment that surfaces ten minutes before the meeting ends and reshapes the planning conversation. Meeting AI captures what the microphone hears; speaker diarization may attribute the comment correctly, but if the spouse is genuinely off-mic, the comment is silent in the transcript. The compensation is the advisor's discipline to either explicitly bring such comments into the on-mic conversation ("Sarah, I want to come back to what you just mentioned about your mom — can you tell me a bit more?") or to make a note in the advisor's own follow-up that captures the off-record observation. The judgment call is whether the off-record observation belongs in the Smarsh-archived note (where it becomes a recordable communication) or in the advisor's private working notes; the firm's WSPs should address this.

Advisor Body Language and Client Body Language

The third failure is body language and emotional register. The pause that says the client is uncomfortable with the recommendation but doesn't want to push back. The advisor's own moment of recognition that the client's response is misaligned with the prior meeting's positioning. The spouse's eye-roll that telegraphs a household disagreement that hasn't surfaced. Meeting AI captures sentiment to some degree — Sybill's primary differentiator — but the nuanced read of a 35-year client relationship is human work. The compensation is the advisor's narrative note appended to the AI summary, capturing the relational observations that drove the actual recommendation framing.

Document-on-Screen Review Without Verbal Description

The fourth failure is the on-screen document review where the advisor and client are looking at the same Holistiplan summary or RightCapital plan and discussing it without verbalizing the specific lines they are pointing at. The transcript captures "and you can see here that the projected RMD in 2031 is about $42,000 which is what triggers the IRMAA cliff..." but doesn't capture which page or section the advisor and client were looking at. The compensation is the advisor's discipline to verbalize the specific reference ("on the Holistiplan summary, line 6 of the RMD projection table...") and to attach the actual document to the meeting record.

Hallucination in the AI-Generated Summary Itself

The fifth failure is the one the L1 Ch2.2 hallucination lesson catalogs — the AI summary that confidently records something neither speaker actually said, the action-item list that includes a step the client didn't agree to, the compliance flag that misclassifies a routine statement. The compensation is the Cardinal Rule from L1 Ch2.3 — every meeting summary gets the source-system / regulatory / client-specific verification pass before the downstream routing executes. The advisor's 90-second scan of the summary is the operational form of the tier-one source-system check; the principal-review queue under FINRA Rule 2210 catches what the advisor misses.

FINRA Rule 4511 and the Archive Architecture

The supervisory and recordkeeping implications of meeting AI are not optional, and the 2026 enforcement landscape has been emphatic that the post-meeting archive is the artifact most likely to surface in an exam. FINRA Rule 4511 (along with SEC Rule 204-2 for the IA side) imposes the retention obligation on books and records "made and preserved in conformity with the requirements" of the federal securities laws — and the 2026 reading is that an AI-generated meeting transcript and summary, when used as part of an advisor's recommendation or service workflow, is a retained communication. The Smarsh and Global Relay platforms have built explicit AI-capture integrations for the named meeting-AI products to address this; the L3 Ch10.1 lesson develops the end-to-end Zocks-to-Wealthbox-to-Smarsh pipeline; the L4 Ch3 chapter develops the WSP language for the supervisory layer.

The FINRA 2026 Annual Regulatory Oversight Report's framing of GenAI under Rule 3110 reasonable-design supervisory obligations applies with full force here. The supervisory architecture must answer: which meetings are captured, what the retention period is, who has access to the archive, how the access log is maintained, what happens when the meeting AI vendor changes (vendor-lock-out risk on the historical archive), and how the principal review under FINRA Rule 2210 inspects the AI-drafted client-facing follow-up before it sends. The 2026 mature practice has these answers documented; the 2023-era "we use Zoom recording sometimes" practice does not, and is one exam letter away from a discovery problem.

The wirehouse pilots at Morgan Stanley and Merrill have absorbed the supervisory architecture from the home-office side — what tools may be used, with what client-NPI handling under Reg S-P 17 CFR Part 248 (and the May 2024 amendments' 30-day breach clock), what archive integration is required, what principal-review queue handles the AI-drafted communications under Rule 2210. The independent RIA channel has had to build the same architecture firm-by-firm; the Zocks enterprise-tier and the Jump enterprise-tier are explicitly designed to support the architecture rather than leave it to the advisor.

Reg S-P, Client NPI, and the Disclosure Conversation

The advisor cannot deploy meeting AI without addressing the Reg S-P 17 CFR Part 248 implications. The meeting transcript captures client NPI — names, family circumstances, financial details, account information, sometimes account numbers verbalized aloud, sometimes Social Security numbers, sometimes health information that would be HIPAA-adjacent if the household care discussion went deep. The May 2024 Reg S-P amendments tightened the obligations: written incident response program, 30-day breach notification for affected individuals on a sensitive-customer-information breach, vendor oversight. The vendor due diligence on the meeting AI provider includes SOC 2 Type II report review, encryption at rest and in transit, data residency, contract terms on audit rights and breach notification SLA, and ADV Part 2A disclosure of the tool's use to clients. The L4 Ch2.2 lesson develops the 40-question vendor due diligence questionnaire; the operational implication for the meeting-AI deployer is that the disclosure happens up-front, ideally in the engagement letter and the ADV Part 2A, and the client is told what tool is in use and what it captures.

The disclosure conversation with the client typically lives in the meeting opener: "I'll be using our firm's meeting AI tool to capture this call so I can give you a clean follow-up. The tool is hosted under our security policies, your information is protected under our SEC privacy obligations, and the transcript is archived in our compliance system. Is that okay with you?" Most clients say yes without hesitation. The client who declines must be honored — the meeting proceeds without AI capture, and the advisor's notes substitute. The firm's WSPs should address the decline-handling workflow.

How the Workflow Actually Reshapes the Advisor Day

The pre-AI Tuesday for the 200-household RIA advisor — alarm at 6 am, in the office by 7:15, six meetings between 9:30 and 4:30, four meetings end with undocumented follow-ups by 5 pm, the Smarsh archive missing two recordings, the Reg BI rollover file missing the documented alternatives — is the canonical productivity crisis the program's Insider Brief opens with. The post-AI Tuesday for the same advisor with Jump or Zocks plugged in looks materially different. Six meetings between 9:30 and 4:30. Each meeting ends with a transcript and summary ready within two minutes. The advisor scans each summary on the way to the next meeting (90 seconds), accepts the action items with minor edits, and the routing executes — Wealthbox logged, follow-up drafted and queued for advisor review, planning software task created, Smarsh archived. By 4:30, all six meetings have documented follow-ups. By 5 pm, the advisor has done the principal review on the AI-drafted follow-up emails and sent them. The Cardinal Rule (L1 Ch2.3) has been applied to each — source-system numbers verified against the planning software and custodian, IRC citations checked, client-specific fit confirmed against the IPS and prior decisions. The Reg BI rollover file from last week now has the documented alternatives because the AI drafted them from the Jump transcript and the advisor verified them.

The advisor recovers the 10+ hours per week the Zocks data describes. The Schwab 2026 study's "AI adoption more than doubled" data point is the diffusion curve catching the rest of the industry. The wirehouse pilots are the channel parallel. The OSJ-aggregator deployment via RFG-and-similar enterprise investments is the BD parallel. Meeting AI is the entry point. The L1 Ch3.2 and L1 Ch3.3 lessons map the planning, tax, estate, CRM, operations, prospecting, and marketing AI categories that compound on top.

Key Takeaways

• AI note-takers ate the advisor calendar in 18 months: Schwab 2026 RIA Benchmarking Study reports adoption more than doubled vs. 2023; Jump and Zocks dominate the Kitces AdvisorTech map's March 2026 meeting-AI cluster; RFG Advisory's enterprise Zocks investment and the Morgan Stanley / Merrill wirehouse pilots signal cross-channel adoption.
• The named five are Jump, Zocks, FinMate AI, Sybill, and Zeplyn, with Pulse360 sitting adjacent in the prep-and-follow-up workflow. Each has a distinct positioning: Jump end-to-end workflow, Zocks compliance-and-archive depth, FinMate AI template control, Sybill sentiment analytics, Zeplyn post-meeting routing.
• Good meeting AI captures six artifacts: verbatim transcript with speaker diarization, structured summary, action items with owner and SLA, next-meeting agenda, compliance flags, and the export package for downstream systems (Wealthbox / Redtail / Salesforce FSC, RightCapital / eMoney / MoneyGuidePro, custodian pre-fill, Smarsh / Global Relay archive under FINRA Rule 4511).
• Meeting AI misses five things: whiteboard math and visual reasoning, off-mic spouse / side-conversation comments, advisor and client body language, document-on-screen review without verbal description, and hallucinated content in the AI-generated summary itself (caught by the Cardinal Rule).
• FINRA Rule 4511 retention applies to AI-generated meeting transcripts and summaries used in advisor workflows. The Smarsh and Global Relay integrations with the named meeting-AI products address the obligation operationally. The FINRA 2026 Annual Regulatory Oversight Report framing of GenAI under Rule 3110 reasonable-design extends the supervisory architecture to the meeting-AI deployment.
• Reg S-P 17 CFR Part 248 (May 2024 amendments) governs the meeting-AI vendor relationship — written IRP, 30-day breach notification, vendor oversight. The disclosure conversation with the client happens up-front (engagement letter, ADV Part 2A, the meeting opener) and the decline-handling workflow lives in the firm's WSPs.
• The post-AI advisor day collapses the 90-minute-to-two-hour downstream cost per meeting by roughly an order of magnitude — the Zocks-published "10+ hours per week saved" figure. The Cardinal Rule from L1 Ch2.3 runs on every meeting summary before the downstream routing executes. The L3 Ch10.1 lesson develops the end-to-end Zocks-to-Wealthbox-to-Smarsh pipeline; the L4 Ch3 chapter develops the supervisory WSP language.