A 78%-probability Monte Carlo plan is the wrong unit to give a client. So is a 92%. So is any percentage by itself. The percentage is a summary statistic over a thousand simulated futures, none of which is actually going to happen — but one of which the household is, in real time, about to live through. The senior retirement-income advisor's craft is to translate that probability surface into a written policy the client can read, hold to, and not abandon at 2:00 p.m. on a peak-VIX Tuesday. The policy has a name, a triggering rule, a numerical action, an annual review cadence, and an IPS-amendment hook. Guyton-Klinger guardrails, the Kitces dynamic-spending ranges, the Bengen 4.7%-revisited framework, and the ratcheting strategies are the four standard options. This lesson installs the AI-assisted workflow that picks among them, generates the household-specific Spending Withdrawal Policy (SWP) statement, the trigger-and-action table, the annual review template, and the IPS amendment language — and ships the package as a Reg BI-documented client deliverable before the next quarterly review.

Why the Static 4% Rule Stopped Being an Answer

William Bengen's 1994 "Determining Withdrawal Rates Using Historical Data" introduced the SAFEMAX — the safe maximum initial withdrawal rate that survived every historical 30-year retirement period in his dataset — and the 4% rule became the most-misused result in personal finance. Bengen himself, by 2023 and again in 2025, had updated the SAFEMAX upward to roughly 4.7% on the back of broader asset-class data, but the 4% reflex outlived the math. The deeper problem with static rules is structural: they ignore the household's actual portfolio path. A retiree who started in 2000 spent the next 14 years burning down a portfolio that never recovered to its starting value in real terms; a retiree who started in 2009 watched their portfolio triple while withdrawing the same nominal dollars and ended with multiples of their starting value. Both followed the same rule. Neither was the same household.

The senior advisor's working unit is not the rule — it is the family's policy, codified as a written statement, attached to the IPS, reviewed annually, and updated against the household's own Monte Carlo position. The policy specifies (1) which rule the household is on, (2) what the trigger conditions are, (3) what the action is when a trigger fires, (4) the annual review cadence, and (5) the IPS-amendment hook when the policy needs to change. That is the deliverable; the AI's job is to draft it accurately, ground every number in the household's own plan, and surface the alternatives considered for the Reg BI file.

The Four Rule Families and When Each Fits

Four rule families dominate the practice in 2026. Each has a published canon, a regulatory-defensible documentation pattern, and a household profile it fits best.

Guyton-Klinger Guardrails

Jonathan Guyton and William Klinger's 2006 paper "Decision Rules and Maximum Initial Withdrawal Rates" introduced four rules: the portfolio-management rule (asset-allocation discipline), the withdrawal rule (annual inflation adjustment subject to constraints), the capital-preservation rule (a 10% withdrawal cut when the current withdrawal rate exceeds the initial rate by more than 20%, with up to 15 years remaining), and the prosperity rule (a 10% withdrawal raise when the current withdrawal rate drops below the initial rate by more than 20%). The framework is bidirectional — cuts in drawdowns, raises in bull runs — and tied to a year-over-year recomputation against the prior-year inflation index. Guyton-Klinger fits the household that wants a written, mechanical policy: the rule fires, the action happens, no ad-hoc judgment. It is the most-cited dynamic-spending framework in retirement-income literature and the most defensible under Reg BI because the rules are explicit and the action is mechanical.

Kitces Dynamic-Spending Ranges

Michael Kitces' formulation, building on the Guyton-Klinger tradition and Jonathan Clements' work on flexible withdrawal, presents the spending ranges differently — a target initial withdrawal rate (often 5.0-5.5% rather than 4%) with explicit upper and lower bounds that trigger flexible spending adjustment. The framework emphasizes the year-by-year decision: how much to adjust, how to think about the floor (essential spending the household cannot cut) versus the ceiling (discretionary spending that flexes), and how the household's actual experience updates the policy over time. Kitces ranges fit the household that wants a flexible, professionally-monitored policy rather than a strict mechanical rule — the household that will tolerate "let's revisit this annually with your advisor" as the policy language.

Bengen 4.7% Revisited

The 2023-2025 Bengen updates — including the EVENESEN broader-asset-class analysis and Bengen's own subsequent publications — produced revised SAFEMAX numbers in the 4.5-4.7% range for 30-year retirement horizons. The static-rule pedigree remains attractive for households who want simplicity and a single number; the updated SAFEMAX is the modern version of that pedigree. Bengen Revisited fits the household whose plan probability is high (95%+ at the new SAFEMAX), whose risk tolerance for spending volatility is low, and who will not engage with mechanical rule-firing year over year.

Ratcheting Strategies

Ratcheting strategies — sometimes called "spending raises" or "smile spending" — adjust withdrawals upward when the portfolio reaches new highs (the "ratchet up") but never downward, with the household's spending trending up in real terms through the early-retirement years and then flattening (the David Blanchett-style retirement-spending smile). The strategy fits the household with a high probability of success at the starting withdrawal rate, a desire to spend more in the active early-retirement years, and an explicit acceptance that the ratchet creates path-dependence in terminal portfolio value. The Kitces ratchet variant introduces specific triggers (a 50% real portfolio gain, e.g.) for the ratchet.

How the AI Picks Among the Four

The AI is not allowed to "choose the rule" — that is a fiduciary judgment the registered advisor owns. The AI's job is to surface the household-specific evidence and produce the comparison the advisor reads to make the choice. The selection workflow has five inputs and four outputs.

Inputs

The five inputs are pulled from the household's planning system and CRM: (1) the current Monte Carlo position from eMoney / RightCapital / MoneyGuidePro, including the probability-of-success surface and the year-by-year withdrawal pattern the plan assumes; (2) the household's stated risk tolerance for spending volatility — the question "are you willing to take a 10% spending cut in a bad market year to preserve plan probability?" — captured in the IPS or the most recent review notes (Wealthbox / Redtail / Salesforce FSC); (3) the floor / ceiling decomposition of household spending — the essential bills (mortgage or rent if any, healthcare premium, food, utilities, property tax) versus the discretionary (travel, dining, gifts) — extracted from the Holistiplan-adjacent budget intake or the household's own statement; (4) the household's risk profile and time horizon — the L3 Ch1 workflow audit and the L2 Ch5 IPS draft supply the inputs; and (5) the household's explicit preference for mechanical-rule policies versus advisor-monitored flexibility.

Outputs

The four outputs are the deliverables: (1) the household's SWP statement — a one-page, plain-English document the household signs, naming the chosen rule, the initial withdrawal rate, the floor and ceiling, the trigger conditions, the action when a trigger fires, the annual review cadence, and the policy-update hook; (2) the trigger-and-action table — a structured matrix the advisor uses at every review to mechanically apply the rule (does today's withdrawal rate exceed initial by 20%? if yes, 10% cut; if no, no action); (3) the annual review template — the structured agenda for the household's annual policy review, including the year's actual experience, the rule's reading on it, the proposed action, and the documented advisor judgment; and (4) the IPS amendment language — the exact paragraph that gets inserted into the household's Investment Policy Statement so the SWP becomes governance, not informal preference.

Three Households, Three Different Policies

The selection workflow earns its keep when the same probability surface produces three meaningfully different policies for three meaningfully different households.

Household A — The Mechanical-Rule Household, 71/69 Retirees, 78% Plan

The Martinezes, $1.1M portfolio, drawing $5,600/month (6.1% initial withdrawal rate), low spending-volatility tolerance, strong preference for written rules that they can read and hold to without depending on advisor judgment. The AI's policy recommendation: Guyton-Klinger guardrails — capital-preservation rule with 10% cut if current WR exceeds initial by 20%, prosperity rule with 10% raise if current WR drops below initial by 20%, both with a maximum cumulative adjustment over five years. The SWP statement says: "If the market drops materially and our annual withdrawal rate climbs above 7.3% (our initial 6.1% plus 20%), we will reduce monthly withdrawals by 10% for one year and revisit at the next annual review. If our withdrawal rate falls below 4.9% (our initial 6.1% less 20%), we will raise monthly withdrawals by 10%. The advisor will compute these annually and explain any action in writing." Reg BI documentation: alternatives considered (Kitces ranges, Bengen Revisited, ratchet), client-specific rationale tied to spending-volatility tolerance.

Household B — The Flexibility Household, 67/65 Retirees, 92% Plan

The Hendersons, $1.8M portfolio, drawing $7,400/month (4.93% initial withdrawal rate), moderate spending-volatility tolerance, a stated preference for "let's revisit this every year with you" rather than a mechanical rule. The AI's policy recommendation: Kitces dynamic-spending ranges with a 5.0% target initial WR, an upper bound at 6.0% (above which discretionary spending steps down) and a lower bound at 4.0% (below which discretionary spending steps up). The SWP statement names the floor (essential spending of $4,800/month not subject to flex) and the ceiling (discretionary $2,600/month subject to annual flex). The IPS amendment ties the policy to the annual review cadence and the planning-system re-run. Reg BI documentation: alternatives considered, client-specific rationale, IPS amendment.

Household C — The Ratchet Household, 64/62 Pre-Retirees Entering Year One, 96% Plan

The Lees, $3.2M portfolio, planning to draw $9,500/month starting next year (3.6% initial withdrawal rate), high probability of success, explicit desire to "spend more while we can travel" in the active early-retirement years. The AI's policy recommendation: Kitces-style ratchet — start at 3.6%, ratchet up by 10% whenever real portfolio value reaches a new high above the starting value by 50%, never ratchet down. The SWP statement includes the smile-spending shape (higher in years 1-10, flattening in years 10-20) and the explicit terminal-value path-dependence acknowledgment. Reg BI documentation: alternatives considered (a higher static rate would be aggressive given the time horizon; Guyton-Klinger guardrails would be unnecessarily mechanical), client-specific rationale tied to stated lifestyle preference and high plan probability.

The Five Deliverables in Detail

Beyond the four standard outputs, the senior advisor's package includes one additional deliverable: the year-one execution memo. Together, the five deliverables ship as the household's withdrawal-policy file.

SWP Statement

One page, plain English, written at an 8th-grade reading level. Names: chosen rule, initial WR, dollar floor of essential spending, dollar ceiling of discretionary spending, trigger conditions in dollar and percentage terms (not just abstract), action when a trigger fires, annual review cadence and date, the planning system the rule is anchored to (eMoney / RightCapital / MoneyGuidePro), and the household's signature line acknowledging the policy. The AI drafts; the advisor reviews; the household signs. The SWP statement is itself a recordable communication under FINRA Rule 2210 (for the BD side) and is retained under FINRA Rule 4511 / SEC Rule 204-2.

Trigger-and-Action Table

A structured matrix the advisor uses at every review. Columns: condition (e.g., "current WR > initial × 1.20"), today's value, today's reading (true / false), required action, dollar impact, advisor judgment-override flag, signoff line. The table is reusable across reviews; the AI populates today's values from the live data feed.

Annual Review Template

A structured agenda for the annual policy review: (1) the year's actual experience (return, withdrawal, year-end value), (2) the rule's reading on it, (3) any trigger that fired, (4) the action the rule requires, (5) any advisor judgment-override and its rationale, (6) the household's discussion of the policy, (7) any amendment proposed, and (8) the next review date. The template is itself a Reg BI artifact when an action is recommended.

IPS Amendment Language

The exact paragraph that gets inserted into the household's IPS. Names the rule, the initial WR, the trigger conditions, the action, the review cadence, and the change-process (advisor proposal + client signature). The IPS-amendment hook makes the SWP governance rather than an informal advisor preference. Stored in the firm's document vault (Wealth.com for estate-adjacent firms; the firm's own drive otherwise).

Year-One Execution Memo

The Reg BI memo that documents the initial recommendation: the household profile, the four rules considered with their respective fit assessments, the chosen rule, the alternatives considered, the client-specific rationale, the costs (any planning-fee impact), and the registered person's signoff. The memo is the file that defends the recommendation in a 2026 SEC exam, a 2026 FINRA AWC-pattern review, or a client claim.

The Regulatory Spine and the Tool Stack

The SWP package is anchored in the same regulatory regimes as every other AI-assisted recommendation workflow in this program. Reg BI §240.15l-1 attaches: the SWP recommendation is a recommendation about an investment strategy (withdrawal/decumulation), the four obligations (Disclosure, Care, Conflict, Compliance) apply, the alternatives-considered file is required, and the registered person owns the signoff. Marketing Rule 206(4)-1 attaches when the SWP statement is templated or repurposed; the January 2026 SEC staff FAQs govern. FINRA Rules 2210, 3110, and 4511 apply: the SWP statement and the annual review memos are communications and books-and-records; principal review under Rule 2210 and supervisory architecture under Rule 3110 (the FINRA 2026 Annual Regulatory Oversight Report's agentic-AI-adjacent framing) govern. Reg S-P 17 CFR Part 248 May 2024 amendments apply to the NPI flowing through the workflow; the tenant-isolated LLM, the IRP, the 30-day breach notification readiness, and the vendor oversight stack are non-negotiable. NY DFS 23 NYCRR 500 layers on for NY-domiciled firms with the 72-hour cyber event notification.

Tool stack: eMoney / RightCapital / MoneyGuidePro for the Monte Carlo surface and probability re-run. Holistiplan for the federal marginal and AGI projection if the SWP interacts with Roth conversion timing (it usually does — see L3 Ch2). Wealthbox / Redtail / Salesforce FSC for the household record and the IPS pointer. Wealth.com or the firm's document vault for the IPS itself. The LLM layer (Microsoft Copilot, Claude through a tenant-isolated wrapper, OpenAI Enterprise, or Salesforce Einstein for FSC) for drafting. Smarsh / Global Relay for the archive. Orion Eclipse or BlackRock Aladdin Wealth for the live portfolio value the rule fires against.

What This Workflow Replaces, and What It Cannot

The workflow replaces the "we'll figure it out as we go" informality that defines most households' withdrawal policy in 2026. It replaces the advisor's reflex to apply a generic 4% rule regardless of household fit. It replaces the year-after-year ad-hoc "let me check the plan and get back to you" pattern that erodes trust precisely when trust is needed. It collapses what used to be a multi-hour bespoke policy drafting per household into a 30-minute AI-drafted draft the advisor reviews, the household signs, and the firm archives.

What it cannot replace is the senior advisor's judgment on household fit — the family's actual relationship to spending, the unspoken anxiety about leaving the surviving spouse without enough, the off-the-record preference for "we don't want to be the household that cuts spending the year our daughter has a baby." The AI surfaces the standard policies and the standard rationale; the advisor's read against the household they actually know produces the recommendation that ships. The Cardinal Rule (L1 Ch2.3) — source-system verification, regulatory verification, client-fit verification — runs against every line of the SWP statement before any household signs.

Key Takeaways

• The Monte Carlo percentage is the wrong unit to give a client. The senior advisor's deliverable is a written Spending Withdrawal Policy (SWP) anchored to the household's actual probability surface and signed by the household.
• Four rule families dominate 2026 practice: Guyton-Klinger guardrails (mechanical, bidirectional, capital-preservation + prosperity rules), Kitces dynamic-spending ranges (flexible, advisor-monitored, floor / ceiling), Bengen 4.7% revisited (static, simple, fits low-volatility-tolerance households), and ratcheting strategies (upward only, fits high-probability early-retirement households with stated lifestyle preferences).
• The AI's job is to surface the household-specific evidence, not to pick the rule. Five inputs (Monte Carlo position, risk tolerance for spending volatility, floor/ceiling decomposition, household risk profile + horizon, preference for mechanical vs. flexible) drive the comparison; the registered advisor's signoff owns the choice.
• The package has five deliverables: SWP statement (one page, plain English, household-signed), trigger-and-action table (structured matrix for every review), annual review template, IPS amendment language (the SWP becomes governance), and the year-one Reg BI execution memo.
• Three households produce three different policies: the Martinezes get Guyton-Klinger guardrails (mechanical, written); the Hendersons get Kitces dynamic-spending ranges (flexible, floor/ceiling); the Lees get a Kitces-style ratchet (upward only, smile-spending shape).
• The regulatory spine is the standard stack: Reg BI §240.15l-1 four obligations, Marketing Rule 206(4)-1 on reuse, FINRA Rules 2210/3110/4511 + Reg Notice 24-09 + FINRA 2026 Annual Regulatory Oversight Report, Reg S-P 17 CFR Part 248 May 2024 amendments, NY DFS 23 NYCRR 500 for NY firms.
• The SWP intersects with adjacent senior-advisor lessons: it consumes the drawdown coaching protocol (L5 Ch7 L1) at every drawdown, the TEY workflow (L5 Ch7 L2) on the income side, the Roth conversion ladder (L3 Ch2), the RMD calendar (L3 Ch3), and the IRMAA two-year lookback (L3 Ch4) — the senior advisor's craft is integrating them.