A practice owner who buys Jump, Zocks, Holistiplan, and Wealth.com on the same Monday and expects the team to be running the integrated pipeline by Friday is going to spend the next ninety days watching that thesis collapse. Real advisor AI adoption is a measurable, predictable curve with named failure modes, a 10% early-adopter cohort that does most of the pulling, and a training design problem that gets harder the more generations sit in the meeting. This lesson installs the 90-day adoption playbook the practice owner, COO, or CCO uses to move a 5-to-50-advisor team from "we just licensed it" to "this is how we work" without burning the champions, alienating the late majority, or producing the kind of half-deployed compliance gap the May 2026 FINRA AWC pattern punishes.

Why Most Advisor AI Deployments Stall at Week Six

Schwab's 2026 RIA Benchmarking Study reports adoption of AI tools has more than doubled vs. 2023, with Jump and Zocks dominating the meeting-AI category and Holistiplan and FP Alpha dominating the planning-extraction category. What that headline hides is the second-derivative finding the same study buries on page 41: among firms that licensed an AI tool in the prior 12 months, roughly a third of advisor seats were not actively using the tool by month three. The license was active. The training had been delivered. The vendor onboarding email had been read. The advisor still wasn't using it.

The pattern is consistent across the dozen named tools in the AdvisorTech stack — Jump, Zocks, FinMate AI, Sybill, Zeplyn, Holistiplan, FP Alpha, Wealth.com, fpPathfinder, RightCapital, eMoney, MoneyGuidePro, Orion Eclipse, 55ip, Wealthbox, Redtail Engage, Salesforce Financial Services Cloud with Einstein, Practifi, Smarsh, Global Relay, Catchlight, SmartAsset, and Microsoft Copilot. The stall is not a tool problem. The stall is a change-management problem with four named failure modes that show up at predictable points in the curve: advisor resistance ("I already have a workflow"), compliance fear ("the CCO hasn't blessed it"), integration friction ("it doesn't load cleanly into Wealthbox"), and the gravitational pull of the unmeasured baseline ("my current way is fine"). A practice that doesn't plan for each by week — and doesn't equip a named internal champion to break each — will land at the same one-third dead-license rate the Schwab data describes.

The 90-Day Adoption Curve, by Week

The curve has five phases. The week-by-week framing is taken from the operating playbooks of three named-named-advisor deployments — RFG Advisory's enterprise Zocks rollout, the Holistiplan 2026 Enterprise Advisory Board firms' standard onboarding, and the wirehouse pilot patterns Morgan Stanley and Merrill have published in 2025-2026 trade press.

Weeks 1-2 — Foundation

The CCO signs the vendor due-diligence file (SOC 2 Type II reviewed, Reg S-P 17 CFR Part 248 vendor oversight checked, GLBA Safeguards alignment confirmed, contract executed). The ADV Part 2A amendment is queued if the tool processes NPI in a new way. The WSPs are updated under FINRA Rule 3110 reasonable design. The Smarsh or Global Relay archive connector is configured and tested. Two named users — the internal champion (see next section) and the head advisor — get the tool first, run it on three real client artifacts (a Zocks transcript, a Holistiplan 1040 scan, a Wealth.com trust intake), and produce the firm's first templated Reg BI memo, follow-up email, and IPS draft using the tool. Outcome of weeks 1-2: a working pipeline, two trained users, an archived test set, and the in-house "this is how it works here" examples that the rest of the team will copy. Skip this phase and the next ten weeks deliver chaos.

Weeks 3-4 — Pilot Cohort

Three to seven advisors join (the 10% early-adopter cohort plus the most curious of the early majority). They use the tool on their own books with the champion shadow-supporting them. Each pilot advisor runs the tool through the four highest-leverage workflows the practice owns — meeting prep, post-meeting follow-up, IPS update, Reg BI memo drafting — and produces a one-page artifact set the rest of the team will reference. Compliance reviewer (CCO or designated principal) reviews the pilot output under Rule 2210 and Marketing Rule 206(4)-1, locks the "this is the firm's voice" templates, and approves the first batch of system prompts for the firm prompt library. Outcome of weeks 3-4: 5-9 active users, a small library of firm-approved templates, and the first round of compliance-graded edits captured for the prompt library.

Weeks 5-8 — The Resistance Window

This is the make-or-break window. The early adopters are bored — they've moved on to the next tool. The early majority is testing the water. The late majority is asking "why are we doing this?" and starting to mention vendor pitches they saw at the last conference for the competing product. The first integration friction surfaces — Wealthbox custom fields don't load, the Holistiplan-to-RightCapital handoff produces a stale plan, the Zocks-to-Smarsh archive missed two recordings. The champion's job in weeks 5-8 is to surface every friction publicly, fix the top five, and prevent the resistance from coalescing into a "this isn't working" narrative that the late majority will use as cover. Outcome of weeks 5-8: ~50% of seats active, the top friction items resolved, and the first ROI metrics (hours recovered, NIGO rate, meeting-to-follow-up SLA) showing measurable lift.

Weeks 9-10 — Momentum

The early majority is in. The late majority is watching the metrics from weeks 5-8 and starting to convert. The CCO publishes the first "AI in the practice" supervisory metrics report under Rule 3110. The advisor pod restructuring conversation starts — paraplanners are getting their hours back, associate advisors are doing more meeting prep, the senior advisor is closing more loops per week. The L2 advisor library (the 25-prompt set built in the L2 capstone) is fully in production. Outcome of weeks 9-10: ~75% of seats active; the firm-approved prompt library at v2; the first month's ROI dashboard published to the partners.

Weeks 11-13 — Institutionalize

The tool stops being "the AI rollout" and becomes "how we work." The last 25% of seats convert (the laggards and the deferred-by-vacation cohort). The training program for new hires now includes the tool from day one. The annual ADV update reflects the tool. The Smarsh archive has 90 days of clean capture. The Reg BI files for the quarter reference AI-generated drafts plus documented human review consistently. The metrics that matter (next lesson: hours recovered, households per advisor, meeting-to-follow-up SLA, NIGO rate, close rate) hit the targets the practice set in week 1. Outcome of weeks 11-13: 90%+ adoption, institutional muscle memory, and a Smarsh-archived record set that survives a 2026 SEC exam.

The Four Known Failure Modes and the Named Mitigations

Each of the four failure modes has a 2026-tested mitigation. The mitigation is owned by a named role, not by "the team."

Advisor Resistance

The senior advisor with a 20-year workflow looks at Jump or Zocks and asks "what does this do that my paraplanner doesn't?" The answer "saves you 10+ hours a week per the Zocks data" doesn't land — it sounds like vendor marketing. The mitigation is the side-by-side prep race: the champion runs the AI prep workflow on three of the resistant advisor's actual upcoming meetings, side-by-side with the advisor's manual prep, and lets the time delta speak. The Schwab 2026 study and the Kitces time-and-task research (roughly 36% of advisor hours in meeting prep and servicing) give the framing; the live demo on the advisor's own clients delivers the conversion. The named owner: the internal champion plus the head advisor as co-sponsors. The CCO is explicitly not the lead here — compliance leading the conversion is the failure pattern.

Compliance Fear

The advisor reads about the Delphia and Global Predictions AI-washing settlements, the FINRA 2026 Annual Regulatory Oversight Report's section on agentic AI under Rule 3110, the Reg S-P 17 CFR Part 248 May 2024 amendments and the 30-day breach clock, and concludes the safest thing to do is not use the tool. The mitigation: the CCO publishes the practice's written AI Use Policy (the L1 capstone deliverable), the WSPs under Rule 3110, the pre-use review protocol under Rule 2210 and the Marketing Rule, the Smarsh archive policy under Rule 4511 and SEC Rule 204-2, and a one-page "what is approved, what is prohibited, what triggers principal review" decision tree. The named owner: the CCO. The framing that converts: "We have written policy. You're more exposed running the tool informally than running it inside the policy."

Integration Friction

The Zocks-to-Wealthbox handoff drops the custom field. The Holistiplan extraction loads into RightCapital with last year's bracket data. The Jump-to-Salesforce Financial Services Cloud sync requires a re-auth every 48 hours. Each friction item by itself is fixable; the cumulative effect is the late majority's "see, it doesn't work" narrative. The mitigation: a named integration owner (typically the ops lead or an associate advisor with technical fluency) whose week-5 deliverable is the top-10 friction list with named vendor tickets, named workaround documentation, and named "this is fixed" dates. The CRMs (Wealthbox, Redtail Engage, Salesforce FSC + Einstein, Practifi) and the AI tools coordinate in 2026; the friction is real but not novel, and the vendor support teams have playbooks. Named owner: the ops lead.

The Unmeasured Baseline

The most insidious failure mode is the absence of pre-deployment metrics. If the practice has never measured advisor hours, meeting-to-follow-up SLA, NIGO rate, or close rate, the post-deployment ROI conversation reduces to anecdote. The mitigation: weeks 1-2 include a baseline measurement sprint. The practice's existing time-tracking (HoursTrackable, Toggl, or the time fields inside Wealthbox / Redtail / Salesforce FSC) captures a two-week baseline before pilot users start. The CRM activity exports give the baseline meeting-to-follow-up SLA. The custodian portal (Schwab Advisor Services, Fidelity Wealthscape, Pershing NetX360+, BNY Mellon) gives the baseline NIGO rate. Named owner: the ops lead with the COO or practice owner as sponsor. The L4 Ch5 L2 ROI dashboard lesson (next) is the operational form of this discipline.

The Internal-Champion Pattern

The single most-cited variable across advisor AI deployments in 2026 is the internal champion. Schwab 2026 and the trade press both name it. The pattern is consistent enough to encode as a role.

The internal champion is not the practice owner, not the CCO, not the CTO if the firm has one, and not the smartest AI user the firm employs. The champion is the most-respected workflow operator on the team — typically a senior associate advisor or a head paraplanner — who is curious about AI, runs their own workflow well, has trust capital with the rest of the team, and has the bandwidth to spend 10-15 hours a week for 90 days on the rollout. The champion's job is fivefold: (1) be the first non-leadership user of the tool and produce the firm's example artifact set; (2) hold weekly 30-minute office hours where any advisor can bring a real client artifact and get a live demo; (3) run the friction list (weeks 5-8) and surface every issue to the ops lead and the vendor; (4) draft and refine the firm prompt library with CCO review; (5) report weekly to the practice owner and CCO on adoption metrics, with the unfiltered version of where each advisor sits on the curve. The champion is not a free role. The job has to be compensated — typically a one-time 90-day stipend ($5,000-$15,000 depending on practice size) plus a quarterly bonus tied to adoption metrics. Practices that try to add the champion job to a full-time paraplanner's existing load watch the champion burn out by week six and the deployment stall.

How to Pick the Champion

Three criteria, in order. (1) Trust capital with the rest of the team — does the champion already have a track record of helping other advisors solve workflow problems? (2) Curiosity about AI specifically — has the champion already tried ChatGPT, Claude, Microsoft Copilot, or a vertical wealth tool on their own initiative? (3) Workflow strength — does the champion already run a clean, organized workflow that other advisors copy? An advisor who scores high on AI curiosity but low on trust capital is the wrong pick — the rest of the team will discount the demos. An advisor who scores high on trust but won't experiment with the tool is the wrong pick — the demos won't get built. The named-named pattern: in a 15-advisor firm, the right champion is usually obvious within two questions of asking the existing advisors "who do you go to when your workflow is stuck?"

Training Design for Multi-Generational Advisor Teams

A 50-year-old senior advisor with a 25-year book, a 38-year-old G2 advisor with a CFP and seven years of practice, a 28-year-old associate advisor two years out of training, and a 22-year-old client service associate fresh out of school all need different training designs for the same tool. The 90-day curve is the same for all four; the on-ramp differs.

The 50-year-old senior advisor learns best through one-on-one live demos on the advisor's own clients, run by the champion. The cognitive load of "here is a new tool" is high; the cognitive load of "let me show you what it does for the Hendersons" is low. The 38-year-old G2 advisor learns best through structured cohort training — three to five advisors in a room, working through the firm's standard workflows together, with the champion facilitating. The 28-year-old associate learns best through documentation and self-serve — the firm prompt library, the vendor's training portal (Zocks Academy, Jump University, Holistiplan training videos, FP Alpha onboarding), and an unstructured Slack channel where questions get answered. The 22-year-old CSA needs the workflow embedded in their first 30 days of onboarding so the tool is never "new" — it is just part of how the firm works.

The training calendar should reflect this. Weeks 3-4 (pilot cohort): one-on-one with the senior advisor pilots, structured cohorts for the G2 advisors. Weeks 5-8 (resistance window): self-serve for the associates and CSAs, continued one-on-one and cohort for the seniors and G2s. Weeks 9-10 (momentum): structured cohort for the laggards. Weeks 11-13 (institutionalize): the training is now part of new-hire onboarding and the role description for every position. The L5 Ch4 lesson on building an AI-literate workforce at scale extends this pattern from a single practice to an enterprise. The L2 capstone (25-prompt advisor library) and L3 capstone (workflow playbook) are the artifacts the training delivers against.

What the CCO Tracks During the 90 Days

The CCO is not the deployment lead, but the CCO owns three deliverables that have to land before day 90. First, the written AI Use Policy and the WSPs under FINRA Rule 3110 — distributed in week 1, reviewed and signed by every advisor by week 4, refreshed by week 13 with lessons learned. Second, the principal review queue under FINRA Rule 2210 and Marketing Rule 206(4)-1 — staffed and operating by week 3, sampling at a documented rate by week 8, and producing an exception log by week 13. Third, the archive coverage check under Rule 4511 and SEC Rule 204-2 — every Zocks/Jump recording, every meeting summary, every advisor edit, every supervisor signoff, every AI-generated client communication archived in Smarsh or Global Relay by week 4 (the late capture window is week 1-2 baseline + week 3 ramp). The CCO publishes a weekly archive-coverage report; gaps trigger same-week remediation.

The CCO also flags two things the deployment lead might miss. (1) The Reg S-P incident response trigger — if a pilot advisor accidentally pastes client NPI into a non-approved tool (Microsoft Copilot personal, ChatGPT free, Claude Pro personal), the 30-day breach clock under the May 2024 amendments may have started; the IR plan kicks in. (2) The ADV Part 2A trigger — if the new tool processes NPI in a way the prior ADV did not disclose, the off-cycle amendment is required. The L4 Ch7 L2 lesson on ADV strategy is where this becomes routine; during the 90-day curve, the CCO catches it manually.

The End-of-Day-90 Checklist

Day 90 produces a checklist the practice owner runs with the COO, the CCO, and the champion. Twelve items: (1) license utilization ≥90% of paid seats active in the prior 14 days; (2) firm prompt library at version 2 or higher; (3) meeting-to-follow-up SLA improved by at least 30% vs. baseline; (4) NIGO rate improved or steady (not worse — NIGO is the leading indicator of operational AI maturity, see L4 Ch5 L2); (5) Smarsh / Global Relay archive coverage ≥99% on AI-touched artifacts; (6) WSPs under FINRA Rule 3110 updated and signed; (7) ADV Part 2A amendment filed if NPI handling changed; (8) ROI dashboard live, showing hours recovered, households per advisor, AUM/advisor; (9) champion stipend paid; (10) any Reg S-P incidents documented and closed; (11) Marketing Rule 206(4)-1 review of any AI-generated client-facing content produced in the 90 days; (12) the day-91 maintenance plan — who owns continued training, prompt library updates, friction reporting, and vendor relationship.

If 9 or more of the 12 are green, the deployment is institutionalized and the practice moves to maintenance mode. If 4-8 are green, the practice extends the rollout 30 days, names the gaps explicitly, and tightens the failing items. If fewer than 4 are green, the deployment has stalled — usually because the champion role was unfunded, the CCO was overweight as deployment lead, or the baseline metrics were never captured. The L4 Ch5 L2 ROI dashboard lesson is the operational follow-on; the L4 Ch6 L1 risk register lesson handles the residual risk items the deployment surfaces.

Key Takeaways

• The 90-day curve has five phases. Weeks 1-2 foundation, weeks 3-4 pilot cohort, weeks 5-8 the resistance window, weeks 9-10 momentum, weeks 11-13 institutionalize. Each phase has named owners, deliverables, and outcomes.
• Four failure modes, four named mitigations. Advisor resistance (champion + head advisor side-by-side demo), compliance fear (CCO published written AI Use Policy + WSPs under FINRA Rule 3110), integration friction (ops lead with top-10 friction list and vendor tickets), unmeasured baseline (weeks 1-2 measurement sprint before pilot starts).
• The internal champion is the single most-cited adoption variable. Not the practice owner, not the CCO. The most-respected workflow operator, compensated with a 90-day stipend ($5K-$15K) and a quarterly bonus tied to adoption metrics. Picked on three criteria: trust capital, AI curiosity, workflow strength.
• Multi-generational training requires different on-ramps. Senior advisors learn through one-on-one demos on their own clients; G2 advisors through cohort training; associates through self-serve documentation and Slack; new hires through onboarding embed. The curve is the same; the on-ramp differs.
• CCO owns three deliverables. Written AI Use Policy + WSPs under FINRA Rule 3110, principal review queue under Rule 2210 + Marketing Rule 206(4)-1, archive coverage check under Rule 4511 + SEC Rule 204-2. Plus Reg S-P 17 CFR Part 248 (May 2024 amendments) incident-response trigger awareness and the ADV Part 2A amendment trigger.
• Day 90 checklist has 12 items. License utilization, prompt library version, meeting-to-follow-up SLA, NIGO rate, archive coverage, WSPs, ADV amendment, ROI dashboard, champion stipend, Reg S-P incidents, Marketing Rule review, and the day-91 maintenance plan. 9+ green = institutionalized; 4-8 = extend 30 days; under 4 = the deployment stalled.
• NIGO rate is the leading indicator of operational AI maturity. The L4 Ch5 L2 ROI dashboard lesson develops this and the other four metrics. The L4 Ch6 L1 risk register handles residual risk items the deployment surfaces.