The discovery meeting is the highest-stakes hour in an advisor's year. The prospect is deciding fit; the advisor is mapping fact-pattern to planning; both are evaluating each other. The advisor who tries to take longhand notes during this hour either misses the conversational signals that close the engagement or captures so little that the post-meeting reconstruction takes longer than the meeting itself. Jump and Zocks — the two AI meeting platforms that dominate the Kitces AdvisorTech March 2026 map, with adoption that more than doubled vs. 2023 per the Schwab 2026 RIA Benchmarking Study — solve this. But they solve it only if configured correctly for the discovery case: the right fields captured, the right action-item flags, the right handling for a spouse speaking over the prospect, the right compliance-review marker on a sensitive moment, the right Smarsh archive hand-off under FINRA Rule 4511. This lesson installs the discovery-meeting configuration playbook end-to-end.

Why Discovery Is Different From Every Other Meeting Type

A standard quarterly review meeting has a known agenda, a known household, a known portfolio, and a known set of CRM custom fields the meeting-AI populates. The discovery meeting has none of those. The prospect's name is in Wealthbox or Salesforce FSC because Catchlight or a referral source added it last week, but every field below the name is blank. The conversation is wide-ranging — career, family, money story, prior advisor experience, risk tolerance, goals, philanthropic intent, business interests — and the prospect will give the most information in the first eleven minutes when they don't yet know which thread the advisor will pull. The meeting AI needs to capture all of it, structure it for downstream IPS drafting (L2 Ch5) and onboarding (L2 Ch7), and produce a transcript that survives a compliance review and a FINRA Rule 4511 retention obligation.

Configuration choices made before the meeting determine 90% of the post-meeting workflow time. A Jump or Zocks instance configured well for discovery produces a transcript, an AI summary, an action item list, and a CRM field-by-field update push, all retrievable from Wealthbox or Salesforce FSC within 90 seconds of the meeting end. Configured poorly, it produces a wall of unstructured transcript that requires forty-five minutes of advisor cleanup. The difference is configuration.

Jump vs Zocks vs FinMate AI vs Sybill vs Zeplyn — The Discovery-Specific Comparison

The five platforms in the meeting-AI category by May 2026 each have a discovery posture. Jump and Zocks lead by ARR and advisor adoption per the Schwab 2026 study and the Kitces map. FinMate AI is the wirehouse-friendly option with stronger Salesforce FSC integration. Sybill positions on conversational analytics (talk ratio, sentiment, engagement). Zeplyn focuses on structured-data extraction for advisor workflow templates.

For discovery: Jump ships with a discovery-meeting template that maps onto Wealthbox / Redtail / Salesforce custom fields out of the box and produces an IPS-draft-ready summary. Zocks offers more flexible custom workflows and is the choice for firms with bespoke prep packs, plus the RFG Advisory enterprise reference for the larger RIA / OSJ case. FinMate is the path of least resistance for Salesforce FSC shops at wirehouses. Sybill adds the conversational layer — useful in discovery for "did we let the prospect talk?" diagnostics. Zeplyn is the structured-extract choice when downstream destinations are highly templated (Practifi custom forms, Pulse360 client deliverables).

The configuration discipline below uses Jump as the worked example because of its discovery-template default, but every paragraph maps to the Zocks equivalent. The advisor running a different platform substitutes the named field; the workflow shape is identical.

Pre-Meeting Configuration — The Twelve-Field Discovery Template

The discovery meeting template needs twelve structured fields the meeting-AI populates from the conversation. These twelve become the seed of the Wealthbox or Salesforce FSC household record, the input to the L2 Ch5 same-day follow-up draft, and the foundation for L2 Ch5 IPS drafting and L2 Ch7 onboarding.

Field 1 — Goals (qualitative). What the prospect wants the advisory relationship to accomplish, in their words. Free-text capture; the AI extracts the verbatim phrases plus a one-sentence summary.

Field 2 — Time horizons. Specific dates / ages tied to specific goals (retirement at 65, college funding starting 2028, gifting program to grandchildren).

Field 3 — Money story. How the prospect describes their relationship to money, prior financial decisions (good and bad), prior advisor experience, prior planning artifacts (existing IPS? prior trust documents?). Verbatim capture plus extraction.

Field 4 — Risk tolerance signals. Specific verbatim phrases — "I can't watch 2008 happen again" or "I'm OK with volatility if the long-term return is there" — that anchor the IPS risk tolerance section. Quantitative scoring tools fail here; verbatim language is more diagnostic.

Field 5 — Household composition. Spouse, children (current ages, dependents, special needs flags), grandchildren, parents (elder-care exposure), prior-marriage children. The L3 Ch7 non-standard-family lens lives here.

Field 6 — Income sources and structures. W-2, K-1, RSU vesting, business income, royalty / IP, rental, deferred comp, pension, Social Security claim status. Specific dollar magnitudes where the prospect shares them.

Field 7 — Assets inventory. Account-by-account or category-by-category, with custodian / institution names. Roth, traditional IRA, 401(k) at named employer, joint brokerage, trust accounts, 529, HSA, real estate, business interests, concentrated stock positions.

Field 8 — Liabilities and obligations. Mortgage(s), HELOC, student loans (parent or self), business debt, intra-family loans, support obligations (alimony, child support per QDRO).

Field 9 — Tax structure. Marginal bracket (verbatim if the prospect knows; otherwise extracted from prior 1040 if shared), state of domicile, prior-year refund/owed pattern, CPA / tax preparer name.

Field 10 — Estate posture. Will / trust existence, executor / trustee identities, beneficiary designation status, estate attorney name, prior estate-planning activity, charitable intent.

Field 11 — Insurance posture. Life (type, face amount, term remaining, beneficiary), disability (own-occ, coverage, group vs individual), LTC (premium-paying state, daily benefit, inflation rider), umbrella, P&C overview, annuity holdings (named insurer, surrender schedule, internal expense).

Field 12 — Advisor team and conflicts. CPA, estate attorney, P&C agent, business attorney, prior wealth advisor, banker, business partner, board memberships, conflicts to disclose.

These twelve fields configure as Jump's "Discovery Template" or Zocks' "First Meeting Workflow." Each maps to specific Wealthbox / Salesforce FSC custom fields. Each is populated by the AI's post-meeting extraction. Each is reviewable by the advisor in the post-meeting cleanup pass.

Real-Time Action-Item Flagging During the Meeting

The action item list is the artifact that drives the L2 Ch5 same-day follow-up and the L2 Ch7 onboarding checklist. Both Jump and Zocks support real-time advisor flagging during the meeting — a keyboard shortcut, a mobile app tap, or a verbal cue ("let's note that as an action item") that marks a timestamp the AI then extracts as a discrete action item with owner, due date, and category.

The discovery-specific action item categories the configuration should include: (1) Onboarding — documents to send, ACATs to initiate, Reg BI rollover analysis to run, Form CRS / ADV 2A/B delivery, engagement letter execution. (2) Planning analysis — Roth conversion analysis, NUA decision memo, estate gap audit, beneficiary review. (3) Client deliverable — IPS draft, follow-up email, second-meeting agenda. (4) Third-party coordination — CPA introduction, estate attorney intro, P&C agent intro. (5) Compliance review — anything that requires CCO / principal review before action.

The discipline that wins: the advisor flags during the meeting (one keystroke, no interruption to the conversation) rather than relying on the AI's post-meeting inference. AI-inferred action items have a 15-25% false-positive rate (the AI extracts "send the Roth article" as an action item when the prospect's casual mention was rhetorical, or misses an action because the spoken cue was implicit). Advisor-flagged action items have a near-zero false-positive rate.

Handling a Co-Prospect (Spouse) Speaking Simultaneously

The most common technical failure in discovery meeting capture is two prospects speaking at the same time. The husband and wife both want to answer "tell me about your money story," they talk over each other for forty-five seconds, and the AI transcript collapses both voices into a garbled paragraph attributed to "Speaker 2."

Three configurations close this. First, speaker labeling: configure Jump or Zocks at meeting start with explicit speaker names (Sarah Henderson, Michael Henderson, [Advisor]) rather than "Speaker 1 / Speaker 2 / Speaker 3." Both platforms support pre-meeting speaker pre-registration; the post-meeting transcript then attributes by name. Second, voice-print learning: in firms with repeat clients, both platforms learn each client's voice over time and improve attribution. For first-time discovery, this doesn't help yet — but for the second meeting it will. Third, the conversational discipline: at the start of the meeting, the advisor names the meeting AI in plain English ("By the way, we're using a meeting AI today to take notes — it captures voice from each of us and I'll review the transcript afterwards. Anything either of you says will be on the record for our planning work. Comfortable with that?") and asks each prospect to introduce themselves verbally so the AI hears each voice in the first minute.

The introduction protocol satisfies two purposes: it gets clean voice samples for attribution, and it satisfies the consent requirement under state two-party consent recording laws (California, Florida, Illinois, Massachusetts, Pennsylvania, Washington — among others; the firm's WSPs should map the consent rule to every state where the firm meets prospects).

Marking a Moment for Compliance Review

Discovery meetings sometimes drift into territory that warrants pre-publish compliance review on the follow-up artifacts. The prospect mentions a pending lawsuit. The prospect proposes a transaction that would create a Marketing Rule disclosure risk if mentioned in any case study. The prospect references a competitor's testimonial that the advisor's firm might re-cite. The prospect discloses a circumstance (Form U4 affecting event, possible affiliated-person relationship) that affects the firm's onboarding analysis.

Jump and Zocks both support a "compliance review" flag that marks the timestamp for the CCO / principal review queue under FINRA Rule 2210 (BD principal review of client communications) or the firm's Marketing Rule pre-use review queue (L4 Ch3 develops this at scale). The configuration: a dedicated keystroke or button that the advisor uses without interrupting the conversation; the post-meeting workflow routes the flagged segment plus the follow-up artifact draft to the compliance queue before the advisor can send.

The five compliance-review trigger categories the configuration should cover: (1) any third-party testimonial / endorsement the prospect mentions that the firm might later repurpose, (2) any hypothetical performance discussion that might surface in a case-study draft, (3) any Form U4 disclosable event (regulatory inquiry, lawsuit, customer complaint), (4) any prospect statement that could be misinterpreted as a guarantee or promise, (5) any potential conflict (affiliated person, related-party transaction, pre-existing fee arrangement).

The Smarsh / Global Relay Archive Pickup Under FINRA Rule 4511 and SEC Rule 204-2

Every Jump or Zocks meeting transcript is a recordable communication. FINRA Rule 4511 requires BD retention of books and records; SEC Rule 204-2 requires adviser retention of communications and the records that produced them. The May 2024 Reg S-P amendments add the data-protection layer — the archive itself must satisfy vendor-oversight and breach-notification requirements.

The integration architecture: Jump or Zocks exports each meeting transcript, summary, action item list, and advisor-edit history to the firm's Smarsh or Global Relay archive on a scheduled cadence (typically near-real-time for hybrid practices, daily batch for RIA-only practices). The L3 Ch10.1 lesson develops the Zocks-to-Wealthbox-to-Smarsh pipeline end-to-end; for L2 Ch2.2 the configuration check is that every discovery meeting transcript lands in Smarsh / Global Relay within the firm's WSP-defined window, with tamper-evident timestamps, the meeting metadata (date, attendees, duration), the AI tool name and version, and the advisor's edit log.

Retention period: the longer of SEC Rule 204-2 (five years, first two easily accessible) or FINRA Rule 4511 (three years). The practical 2026 convention is the SEC five-year baseline. The May 2024 Reg S-P amendments do not change retention duration but add the breach-notification, IRP, and vendor-oversight overlay.

The Three-Minute Post-Meeting Cleanup Pass

The discipline that makes the configuration work is the three-minute post-meeting cleanup. Immediately after the meeting (before the next one), the advisor opens the Jump or Zocks meeting view and runs five passes: (1) speaker attribution check — any "Speaker 2" labels relabeled to Sarah / Michael / etc.; (2) action item review — confirm the AI-extracted action items match the advisor's intent, delete false positives, add any missed items; (3) twelve-field population check — confirm each of the twelve discovery fields was populated with the captured content; (4) compliance flag review — confirm any flagged segments routed to the compliance queue; (5) push-to-CRM trigger — fire the integration to Wealthbox or Salesforce FSC. Three minutes; reflexive after a week of practice.

The L2 Ch5 same-day follow-up lesson takes the cleaned transcript and twelve-field record as its input. The L2 Ch7 onboarding checklist lesson takes the action item list. The L3 Ch10.1 archive lesson confirms the Smarsh pickup. The advisor's job ends with the cleanup pass; the downstream workflow is automatic from there.

Key Takeaways

• Discovery configuration is the leverage point — Jump or Zocks configured for discovery with the twelve-field template and the action-item flagging produces a 90-second post-meeting workflow; configured poorly, it produces a 45-minute cleanup.
• The twelve fields: goals, time horizons, money story, risk tolerance signals, household composition, income sources, assets inventory, liabilities, tax structure, estate posture, insurance posture, advisor team and conflicts.
• Real-time advisor flagging of action items via keystroke beats AI inference — 15-25% false positives on inferred items, near-zero on flagged ones.
• Co-prospect handling uses speaker pre-registration, voice-print learning, and a verbal introduction protocol that doubles as state two-party consent compliance for the conversation recording.
• Compliance review flags route sensitive moments to the CCO / principal queue before the follow-up artifact can be sent — Rule 2210 BD principal review, Marketing Rule pre-use review, L4 Ch3 development at scale.
• Smarsh / Global Relay archive pickup under FINRA Rule 4511 and SEC Rule 204-2 lands every transcript, summary, action items, and advisor edit log; retention is the longer of five years (Rule 204-2) or three years (Rule 4511).
• Three-minute cleanup pass (speaker attribution, action item review, twelve-field check, compliance flag review, CRM push) is the only manual work; everything downstream (follow-up email, IPS draft, onboarding) flows automatically.