In-meeting note capture for the recurring review is structurally different from discovery capture (L2 Ch2.2). The household is known. The IPS exists. The prep pack (L2 Ch3.1) has surfaced the three planning conversations. The advisor's job in the meeting is not data collection — it is judgment, decision-making, and client coaching. The meeting AI's job in the meeting is to capture the decisions made, the action items committed, the rationale for each Reg BI recommendation, and the client's verbatim affirmations — and to package those into a Rule 4511-compliant retention bundle the firm can hand to a 2026 SEC examiner in 12 minutes. This lesson installs the compliance-grade in-meeting capture configuration in Jump, Zocks, or FinMate AI — real-time action-item extraction, time-stamped client decision logging, the Smarsh-archivable export, and the five-artifact Rule 4511 retention package (transcript + summary + action items + advisor edits + reviewer signoff).

Why Review Capture Is Structurally Different From Discovery

Discovery is data-rich and decision-light: the prospect describes a fact pattern, the advisor maps it to planning, no transactions occur in the room. Review is the inverse: data is mostly known from the prep pack, but transactional decisions happen live — "yes, queue the Roth conversion at $58K for Q4 to stay below the IRMAA cliff" or "yes, execute the rebalance trades this week" or "yes, sign the engagement to coordinate with the CPA." Each decision in the room is a Reg BI recommendation under §240.15l-1, a potential FINRA Rule 2210 communication-trail artifact, and a Marketing-Rule-adjacent client statement (if the advisor later references the decision in a case study or capability deck).

The capture posture changes accordingly. Discovery prioritizes wide-net fact extraction (the twelve-field record). Review prioritizes decision logging (time-stamped, attributed, with documented rationale) and action-item commitment (with owner and date). The Rule 4511 retention bundle is therefore the central artifact — five components stitched together at meeting end and routed to Smarsh / Global Relay under the L3 Ch10.1 pipeline.

The Five-Artifact Rule 4511 Retention Bundle

The bundle that defends a 2026 SEC or FINRA exam on the question "show me what happened in this meeting and the recommendations that flowed from it" has five components.

Component 1 — Time-Stamped Transcript With Speaker Attribution

The raw conversation, attributed by speaker (Sarah Henderson, Michael Henderson, [Advisor]), time-stamped at every speaker change. Jump and Zocks produce this natively; the cleanup-pass relabeling from L2 Ch2.2 applies. The transcript is the foundational record; everything else derives from it.

Component 2 — AI-Generated Summary

A structured summary of the meeting: which planning conversations were discussed, which decisions were made, which action items were committed. The summary is the digestible record; the transcript is the auditable record. Both are retained. The summary's format should mirror the L2 Ch3.1 prep pack's three-conversation structure so the prep brief and the meeting summary read as one continuous narrative.

Component 3 — Action Items With Owner, Date, and Status

Extracted from the transcript via advisor-flagged keystroke (L2 Ch2.2 discipline) and AI inference. Each item carries: action description, owner (advisor / client / third party), due date, status, and category (onboarding / planning / trade / coordination / compliance review). The action item list feeds the L2 Ch3.3 post-meeting routing workflow.

Component 4 — Advisor Edits and Annotations

The advisor's post-meeting cleanup edits to the summary and action items: corrections, clarifications, attribution fixes, and explicit Reg BI rationale annotations for any recommendation that surfaced in the meeting. Edits are tracked with timestamp and edit author; the L4 Ch3 WSPs require this audit trail.

Component 5 — Reviewer Signoff (Principal Review / CCO Where Applicable)

For a hybrid practice or for high-risk content (rollover Reg BI memo, performance discussion, third-party rating mention), the principal or CCO signoff confirms pre-use review under FINRA Rule 2210. For an RIA-only practice with risk-based sampling under L4 Ch3, the signoff applies to the sampled subset; the L4 Ch3 lesson develops the scaling design.

The five components retained together as a bundle, indexed by client + meeting date, satisfy the Rule 4511 / Rule 204-2 retention obligation and support the supervisory reconstruction a 2026 FINRA exam may request.

Real-Time Action-Item Extraction — The Advisor-Flag-Plus-AI-Inference Pattern

The L2 Ch2.2 lesson established that advisor-flagged action items via keystroke have a near-zero false-positive rate compared with the 15-25% rate on AI-inferred items. The review meeting amplifies this: review meetings produce more action items per minute than discovery meetings because the prep pack's three conversations each typically generate 1-3 action items, and the advisor's spontaneous additions during the meeting add more.

The configuration: a single keystroke (typically Cmd-Shift-A or a similar firm-standard) marks a timestamp; the advisor can optionally speak the action item verbally ("let's note that as an action") to provide audio context the AI uses to extract owner, action, and date; the post-meeting cleanup pass reviews the flagged set plus the AI-inferred set and consolidates.

The five action-item categories from L2 Ch2.2 apply here: Onboarding, Planning analysis, Client deliverable, Third-party coordination, Compliance review. Review meetings add two additional categories: Trade authorization (DocuSign-bound, custodian-bound, often tied to a Reg BI memo) and Plan update (RightCapital / eMoney plan-revision tasks). Both flow to the L2 Ch3.3 post-meeting routing.

Time-Stamped Client Decision Logging

The decision log is the Reg BI-defensive artifact. Every recommendation the advisor makes in the meeting — and every client decision to accept, defer, modify, or reject — gets time-stamped, attributed, and tied to the rationale documented in the meeting. The Care Obligation §240.15l-1(a)(2)(ii) and Disclosure Obligation §240.15l-1(a)(2)(i) live in this log.

Example log entry: "11:24 AM — Advisor recommended Roth conversion of $58,000 to Sarah's IRA before year-end, sized to keep MAGI under $206,000 IRMAA cliff and within 24% bracket; rationale: bracket arbitrage between current 24% marginal and projected 28% post-RMD; alternatives discussed: full $96K bracket-fill (rejected — IRMAA cost), zero conversion (rejected — bracket arbitrage value). 11:26 AM — Sarah accepted; Michael accepted; advisor committed to draft Reg BI rationale memo by Friday and queue conversion in Schwab by November 30."

The log format converts the meeting's conversational flow into a structured record that survives any subsequent inquiry. The L3 Ch10.1 archive pipeline lands the log alongside the transcript and summary. The L4 Ch3 WSPs codify the format firm-wide; the L4 Ch7 substantiation file uses the log as one of the contemporaneous-evidence anchors for any AI capability claim.

Smarsh / Global Relay-Archivable Export

The five-component bundle exports to Smarsh or Global Relay via the firm's integration architecture. The export format is typically a structured PDF (or, in 2026, increasingly a JSON-with-PDF bundle) containing: meeting metadata (date, time, duration, attendees, location/Zoom), the five components in sequence, the firm's audit trail (who exported, when, integrity hash), and the references to upstream artifacts (the prep pack brief, prior meeting bundle, IPS version, AI tool name + version, prompt name + version).

The integration cadence: hybrid practices export near-real-time to satisfy the Rule 4511 BD-side immediacy expectations; RIA-only practices can export daily batch under Rule 204-2's more flexible immediacy framework. The May 2024 Reg S-P amendments add the breach-notification (30-day), written incident response, and vendor-oversight overlays for the archive itself. The L4 Ch6 cybersecurity playbook addresses the architecture; for L2 Ch3.2 the discipline is that every meeting's bundle lands in the archive within the firm's WSP window with tamper-evident timestamps and metadata.

Jump vs Zocks vs FinMate AI — Review-Specific Comparison

The three primary platforms each have review-meeting-specific strengths.

Jump ships with a review-meeting template that maps to Wealthbox / Salesforce FSC household custom fields and produces an IPS-compliance-aware summary. The Jump default flows the prep brief's three conversations into the meeting summary's three sections — operationally seamless if the firm is already on the Jump prep workflow.

Zocks offers more configurable workflows with deeper Reg BI-rationale prompts and the RFG Advisory enterprise reference for larger RIA / OSJ deployments. The Zocks decision-log format is widely cited as the most structured among the three.

FinMate AI wins at wirehouses with Salesforce FSC and Microsoft Teams integration. The home-office governance frameworks at the major wirehouses (Morgan Stanley, Merrill, UBS, Wells Fargo Advisors) typically pre-approve FinMate or have a similar in-house product.

The L4 Ch2 vendor scorecard lesson develops the selection framework; for L2 Ch3.2 the operational point is that whichever platform the firm uses, the five-artifact bundle architecture is the same. The platform supplies the components; the firm supplies the discipline.

The 12-Minute Archive Window From Meeting End

The operational target: 12 minutes from meeting end to fully-archived five-component bundle in Smarsh / Global Relay. The breakdown: 30 seconds for the AI to assemble the summary and action items; 90 seconds for the advisor's cleanup pass (speaker attribution, action item review, decision log verification); 3-5 minutes for any Reg BI rationale annotation on recommendations made in the meeting; 3-5 minutes for principal pre-use review on any flagged content; 1 minute for the export and archive pickup. Total: 12 minutes for a typical review with two or three recommendations; longer (up to 25 minutes) for a meeting with a major rollover Reg BI memo or a complex equity-comp decision.

The 12-minute target matches the L2 Ch2.3 follow-up SLA — same operational discipline, same downstream archive pipeline. A practice that hits the 12-minute archive target across 800 meetings/year produces 600-1,200 hours of recovered advisor time relative to the manual-note baseline and a Rule 4511 / Rule 204-2 archive that withstands examination scrutiny.

The Reg BI Rationale Annotation — Where It Lives, How Long It Takes

The Reg BI rationale annotation is the single most common gap in advisor meeting documentation by May 2026. The 2025-2026 FINRA AWC pattern on inadequate rollover documentation turns on this exact failure: the recommendation memo exists, but the documented consideration of alternatives is thin or missing. The in-meeting annotation closes this — every recommendation gets a one-paragraph rationale captured in the decision log noting the alternatives considered, the client-specific reason for the recommendation, the cost / benefit comparison, and the client's affirmation.

The L2 Ch7.2 lesson on 401(k) / 403(b) rollover Reg BI memos develops the full four-alternative framework (leave in plan, roll to new employer plan, roll to IRA, take cash). For other recommendations — Roth conversion, NUA election, asset reallocation, QCD execution, fee change — the analogous alternatives are documented. The discipline is that the in-meeting annotation captures the rationale while the advisor's memory is fresh; the post-meeting written Reg BI memo expands the annotation into a full document but does not have to start from scratch.

The Locked Prompt for In-Meeting Bundle Assembly

The L2 Ch1.1 anatomy-locked prompt for the bundle assembly:

Role. You are a senior CFP-certificant advisor finalizing the post-meeting bundle. You never invent facts and write in plain English at an eighth-grade reading level.

Context. The attached transcript is from a review meeting today with [Client Name]. Attached: prep pack brief, IPS, prior meeting bundle (for context), action items list (flagged by advisor + AI-inferred), decision log (advisor-marked moments).

Task. Produce the five-artifact bundle: (1) transcript (passthrough with speaker attribution verification), (2) AI summary (sections: planning conversations discussed, decisions made, action items, open items deferred), (3) action items table (action, owner, due date, status, category), (4) advisor edit log placeholder, (5) reviewer signoff placeholder.

Format. Structured PDF-bound output. Each section labeled. Decision log entries time-stamped with 24-hour clock. Action items in table format.

Constraints. (1) Do not invent any decision or recommendation not in the transcript. (2) Every recommendation must carry the rationale paragraph if captured in the decision log; otherwise write [need: advisor to add rationale]. (3) Do not state any specific dollar magnitude for a future trade or plan revision unless captured. (4) Do not use marketing language. (5) For any client affirmation reference, use verbatim if available. (6) Action item categories: Onboarding, Planning analysis, Client deliverable, Third-party coordination, Compliance review, Trade authorization, Plan update. (7) The bundle's metadata block at top must include client identifier, meeting date/time/duration, attendees, AI tool name+version, prompt name+version, prep pack brief reference, prior bundle reference.

Monday Morning Deployment

Implementation: configure Jump, Zocks, or FinMate against the firm's review-meeting template using the twelve fields from L2 Ch2.2 (now populated from prior household records rather than blank as in discovery). Add the seven action-item categories including Trade authorization and Plan update. Add the keystroke for flagging compliance review moments — Cmd-Shift-R or firm-standard. Configure the export to Smarsh / Global Relay per the firm's WSP cadence (near-real-time hybrid, daily batch RIA-only). Run the next five review meetings through the configured workflow. After meeting 1, run the cleanup pass; after meeting 5, the cleanup time approaches 90 seconds and the Reg BI rationale annotation becomes reflexive. The 12-minute archive window becomes routine.

The investment is configuration time once; the return is recovered advisor time across hundreds of meetings per year plus the defensibility evidence base that supports L4 Ch3 WSPs, L4 Ch7 Marketing Rule substantiation, and L4 Ch8 M&A diligence simultaneously. The L3 Ch10.1 archive pipeline lesson develops the firm-scale integration; the L4 Ch5 ROI dashboard tracks the operational metrics (12-minute hit rate, principal review SLA, NIGO rate on trade authorizations) as the practice's compliance-and-productivity scorecard.

Key Takeaways

• Review capture is decision-centric, not data-centric — the prep pack (L2 Ch3.1) already supplied the data. Every in-meeting recommendation is a Reg BI artifact under §240.15l-1.
• The five-artifact Rule 4511 retention bundle: transcript + AI summary + action items + advisor edits + reviewer signoff — indexed by client + meeting date, exported to Smarsh / Global Relay via the L3 Ch10.1 pipeline.
• Time-stamped client decision logging is the Reg BI-defensive artifact: every recommendation + every client decision (accept / defer / modify / reject) + rationale + alternatives considered + client's verbatim affirmation.
• Real-time advisor flagging via keystroke beats AI inference on action items; the seven categories (Onboarding, Planning, Client deliverable, Third-party coordination, Compliance review, Trade authorization, Plan update) cover review-meeting action surface.
• The 12-minute archive window from meeting end matches the L2 Ch2.3 follow-up SLA; total 12 minutes for typical review, up to 25 minutes for rollover Reg BI or complex equity-comp decisions.
• Reg BI rationale annotation captured in-meeting closes the 2025-2026 FINRA AWC failure pattern (inadequate rollover documentation); the L2 Ch7.2 four-alternative framework applies.
• Jump, Zocks, FinMate AI all support the five-artifact architecture; vendor selection per L4 Ch2 scorecard. Archive retention is the longer of SEC Rule 204-2 (5 yrs) or FINRA Rule 4511 (3 yrs), with May 2024 Reg S-P breach / IRP / vendor-oversight overlays.