The definitive plan after red-team review — supersedes conflicts in other files.

26 — Revised Final Plan (Plan of Record)

Prepared 2026-07-22. Figures verified as of this date unless marked otherwise.

This file is the plan of record. It incorporates the red-team review (file 25) and the revision pass that followed. Where this file conflicts with any earlier file (00–24, 27), this file governs — the earlier files have been corrected in place, but if a stale number or claim survived somewhere, resolve it in this file's favor and fix the other file. Prompt 2 reads this file first (MASTER_PROMPT L1089–1096).

1. The recommendation (unchanged in essence, corrected in calibration)

Mitchell builds his first year on:

PRIMARY — Realtor Scenario Desk (Model D). Fast, written, guideline-checked answers to hard financing scenarios for a ~30-agent southwest-suburb pilot list, on a documented SLA: acknowledgment ≤1 business hour; written brief ≤4 business hours, Mon–Fri 8:30–5:30. Positioning: "I'm not asking you to replace your preferred lender. Give me one difficult scenario or one buyer who needs a second look."
SECONDARY — Southland First-Home Lab (Model E) (working name — needs Ready naming approval): local education brand on three unclaimed topics — the DPA stack, Cook-vs-Will tax math, and condo-financing triage (post-change framing). Honestly positioned as a trust-and-asset play with small, slow direct yield in year one — not near-term pipeline.
EXPERIMENT — Model F-lite. One Investor/Self-Employed File Lab content franchise + 2–3 scenario drills/week from week 1; binary scale-or-kill at day 60.
NO Zillow Flex (structural conflict: ZHL contact-rate KPI, steering litigation, ~$2,600 gross per $350k closing). Broker license = literacy and optionality only; never both roles on one transaction in year one.
Mortgage-first. Everything on the NOT-NOW list stays there (files 01, 09 §7).

Geography: Tinley Park + Oak Forest + Orland Park + Lockport; phase 2 at day 90+. Products in mastery order: conventional → FHA → VA → the DPA lane Ready actually supports → bank-statement non-QM (if confirmed).

The honest year-one expectation (re-based, file 22 §6): conservative 1–3 funded loans (income rounds to zero), base 5–6, aggressive ~14 — with ≈ $0 income in months 1–5 even in the base case. First funded loan: month 4 base, month 3 only if lucky, month 7–8 conservative. This is why the runway gate (§3.2 below) exists.

2. What changed after red-team review

Every material change from the revision pass, keyed to file 25's finding numbers:

#	Finding	What changed	Where
C1	Missing deliverables	This file, `START_HERE.md`, and all three `decision_gate/` files now exist.	26, START_HERE, decision_gate/
C2	Financial model contradicted itself (hand-set ramps 2–3× funnel capacity)	Ramps re-based: old "conservative" (6) is now base; old "base" (14) is now aggressive; a true conservative ramp (1–3, effectively $0 year) was added. Revenue table, `MONTH_FIRST_FUNDED`, CSV, and workbook all recomputed. Headline everywhere: base ≈ 5–6, not 14.	22 §6, data/financial_model_inputs.csv, workbook, 01
C3	No personal-runway analysis anywhere	New runway module: named variables (`[MONTHLY_PERSONAL_BURN]`, `[LIQUID_SAVINGS]`, `[OTHER_MONTHLY_INCOME]`), `RUNWAY_MONTHS` formula, and the pre-committed rule: runway < 9 months at true-conservative pace → income bridge defined with the manager before Day 1. Runway is now P0-0 in the questions file, Q14a–14c in the manager agenda, and R11's early-warning signal is finally computable.	22 §6A, 27 P0-0, 20 Block 4, 24 R11, CSV
C4	Scripts asserted unverified capabilities as fact ("reviewed with my team," Saturday SLA)	v0/v1 rule across all scripts and SLA language: v0 (default) claims only what Mitchell controls ("written, guideline-checked"); v1 ("human-reviewed by my team," Saturday hours) unlocks only on written verification of a named reviewer / weekend coverage. Saturday dropped from the published SLA (Saturday = same-day ack, brief next business morning). Pre-outreach checklist added to file 10.	16 (header, #1, #3, #6, O2), 09 §2.1/§2.3/§2.5, 00 B4, 05 §J, 10 item 11, 23 L9
C5	First 30 days hung on a small, inattentive employer approving everything in 72 hrs, no contingency	Approval-starved contingency track: Day-1 fallback (30-min compliance-only video call), a defined no-approval-needed holding pattern (drills, sample-brief portfolio, research briefs, facts library, community presence), and day-7/14/21 escalations. Day 21 with no approved copy = employer-viability signal feeding §4 below. R15 rescoped from content-only to all approvals, raised to H/H.	10 (new section), 11 Phase 1, 12 week-2 header, 24 R15
M1	No employer hedge; IL license renewal (2026-12-31) untracked	Plan B written (§4 below): three walk-away triggers, alternate-sponsor shortlist, portability checklist executed now. New risks R20 (license non-renewal, monitored 2026-11-01) and R21 (employer cannot support plan).	26 §4, 24 R20/R21, 09 §6
M2	Files 11/22/23 described three different businesses	All 30/60/90 targets and KPI bands re-based to the corrected file-22 base case; day-90 gate now uses leading indicators only (closings excluded from gating before month 4); the day-45 offer-redesign rule (09 D2) is the single source of truth for that stop rule; new scorecard rule 11: file 22 wins all future disagreements.	11 (all targets/gates), 23 L5/G2/rule 11
M3	Cold outreach in a costume; warm network treated as a conflict	Warm-first pilot entry: 6–8 slots for Crosstown Hybrid/non-Flex agents via family + family intros elsewhere, entering weeks 1–2 before any cold wave (conflict hygiene kept: Flex-team members still disqualified, RESPA applies to family). Sample-brief-first: every Day-1 email leads with a finished synthetic brief attached — proof, not promise. Model D sustainability rescored 9 → 7 (G 9 → 8) with the ramp-phase caveat stated.	15 §1, 09 §2.6/D1, 12, 08
M4	DNC/SAN dependency could silently kill calling in week 1	SAN question promoted to P0-6; explicit fallback rule (no SAN → unscrubbed cells are NO-CALL; sequence runs email/LinkedIn/office-landline/in-person with targets cut); Day-1 written SAN request with owner + date (file 10 item 9b, Decision Sheet line 22).	27 P0-6, 15 §4, 10 items 9/9b, 20 Q18
M5	Four compliance leaks in scripts/calendars	(1) Script #7's listing-specific one-pager killed — generic own-brand one-pagers only, CCO sign-off required. (2) Forwardable brief summary is now terms-free (conclusion + "full brief attached with disclosures"). (3) Open-house/field "live payment math" removed — approved talk track is process/DPA/tax anatomy + "I'll send the compliant worked example." (4) O6's Cook-DPA figures bracketed [CONFIRM LIVE].	16 #7/O6, 09 §2.3 step 4, 12 wk-3 Sat, 13, 14
M6	Condo asset "beat August 3" was arithmetically dead	Asset reframed post-change: "the rules changed August 3 — here's the new playbook," with the January-2027 15%-reserve rule as the forward deadline. All "act before August 3" urgency scrubbed from scripts, plans, and market research. The reframe is honestly the better hook — most agents haven't felt the change yet.	09 §3.3/E1, 06 §5, 11, 12, 16 #14, 18, 01, 08
M7	SLA pressure would push NPI into AI tools	Mandatory redaction gate inserted as workflow step 1.5: manual normalization onto the intake template (initials/ranges only) before any AI window opens; on the brief SOP checklist, not skippable. R16 likelihood raised L → M.	09 §2.3, 24 R16
M8	DPA lane may reduce to everyone's offer; Lab oversold as pipeline	Honest day-90 content expectation set in writing: 0–2 attributed conversations = normal, not failure; Lab repositioned as referral-trust play in files 01/09. DPA-lane-fails recut sketched (§5 below).	09 E1 item 11, 01, 26 §5
M9	Week-1 social load unexamined; miscalibrated gates prescribed more of the toxin	Half-dose week 2 (waves of 5), week-3 full-dose gate on energy ≥6/10; new KPI L14 (energy at close); low-conversation gate response changed from "add a sprint" to "diagnose energy vs. message first"; burnout guard now also triggers on 2 red L14 weeks.	12, 23 L14/rule 10, 11 day-30 gate
m1–m8	Minors	RE/MAX 10 location fixed (Orland Park) in 21; hardcoded `[DATE_60]` replaced with `[WEEK_1_START_DATE]+60` business-day-adjusted; F-lite drills run from week 1 (file 11 aligned to 09); unsubscribe promise replaced with "reply 'stop' — same day" until tooling exists; week-2 calendar gaps (Wed/Fri 9:00–9:30) filled; Oak Forest data flags already present at reuse site; public business-hours definition unified (Mon–Fri 8:30–5:30, file 09 §2.1 is the single source); file 27's one-pager renumbered to nine asks.	21, 09 §4, 11, 16 #3, 12, 09 §2.1, 27

What did NOT change: the strategy itself. The red team's verdict was that the failures were "failures of calibration and follow-through, not of research or intent" — the Scenario Desk + Lab + F-lite architecture, the Flex "no," the compliance regime, the verification discipline, and the stop-rule culture all survived attack and stand.

3. The two gates that precede Day 1

3.1 The decision gate (process)

Nothing in Prompt 2 builds until decision_gate/decisions_needed_before_build.md is resolved and decision_gate/compliance_approval_checklist.md is signed. See decision_gate/prompt_2_readiness.md.

3.2 The runway gate (money)

Before Day 1, Mitchell computes RUNWAY_MONTHS = [LIQUID_SAVINGS] / ([MONTHLY_PERSONAL_BURN] − [OTHER_MONTHLY_INCOME]) (file 22 §6A; questions in file 27 P0-0).

≥ 9 months: proceed; recompute monthly; R11 alarm at <6.
< 9 months: the plan as designed is not viable on savings alone. Define the income bridge before Day 1 — draw negotiated in writing (file 20 Q14a/14c), a part-time income floor that preserves the 9:30–6:00 calling window, or documented household support. Shrinking runway never authorizes paid leads, over-promising, or Flex.

3.3 The capability gate (claims)

No v1 claim ("human-reviewed by my team," Saturday SLA, any turn-time or speed number) is ever spoken, written, or published until the underlying capability is verified in writing. v0 wording ships on Day 1 regardless — it is fully honest and fully sufficient.

4. Plan B — the employer hedge (red-team M1)

Ready may be the weak link. The plan documents Ready's fragility (R5–R8, R20–R21) and now prepares the move instead of just worrying about it.

Walk-away triggers (written, dated, reviewed at each gate): 1. No written IL product menu by day 30 (P0-2 unanswered despite the meeting + escalations). 2. No IHDA/DPA-pairing answer AND no FHA delegation confirmed by day 60 — the wedge's product spine cannot exist. 3. SLA back end unstaffable by day 30 (no named escalation contact, no reviewer path, no ops answer to Q36) — the flagship promise cannot be operated even at v0. (Related: approved-copy starvation at day 21 — file 10 — counts as a half-trigger and accelerates this review.) 4. License non-renewal signals by 2026-11-15 (R20 monitoring: IDFPR/NMLS checks calendared 2026-11-01 and 2026-12-01).

One trigger = a direct, documented conversation with the manager with a cure period (2 weeks). Two triggers, or one uncured = activate the shortlist.

Alternate IL sponsor shortlist (pre-researched candidates, from file 07's IHDA-participant analysis — all verified IHDA-participating, all active in or near the footprint; diligence questions = the same file-20 agenda): - Neighborhood Loans (IL-based, IHDA participant, strong SW-suburb presence) - Rate (Guaranteed Rate) (Chicago HQ, IHDA participant, large new-LO infrastructure) - Wintrust Mortgage (Chicagoland bank-owned, IHDA participant, community-bank model) - Marquette Bank / Crown Mortgage / Pacor Mortgage (south-side/SW-suburb locals, IHDA participants — smaller, closer cultural fit) A move is a last resort — Mitchell is new and switching sponsors has real costs (relicensing friction, restart of internal credibility) — but a licensed MLO with a growing local asset base must never have zero prepared moves.

Portability checklist — executed NOW, not at crisis (each item is a decision_gate item or a file-20 question): - Personal domain registered and personally owned (file 20 Q19 defines what may point at it). - Self-generated lead/contact export rights in writing (file 20 Q16/Q17, Decision Sheet lines 20–21). - Content copyright position documented (what is Mitchell's vs. Ready's per policy — ask at Q22). - Personal Google Business Profile owned by Mitchell. - The scenario log, drill library, facts library, and pilot-list research live in Mitchell's own accounts (no NPI in any of them, so no GLBA issue in portability).

5. The DPA-lane-fails recut (red-team M8)

If the manager meeting returns "no IHDA path, no Cook-DPA pairing, no FHLB DPP channel," product #4 collapses to seller credits + gift funds — every lender's offer. Pre-committed recut:

Segment priority reorders to: (1) condo/attached-stock triage (the post-08-03 regime is product-agnostic and Ready-independent), (2) Cook-to-Will tax-arbitrage move-up, (3) first-timers via FHA readiness + honest DPA referral (teach the stack, name who offers it, take the trust and the non-DPA files).
Lab Asset 1 (DPA stack) repositions explicitly as a referral-trust play with a stated small expected direct yield — the guide keeps naming IHDA competitors because serving the reader is the moat.
The desk's pitch de-emphasizes DPA and leads with second-look/condo/self-employed lanes.
This recut is a tactical branch. It does NOT by itself trigger Plan B — but combined with trigger 2 in §4 (no FHA delegation), it does.

6. Consolidated 90-day execution view (reconciled numbers)

One business, one set of numbers (file 22 base case; targets are editable assumptions):

Milestone	Day 30	Day 60	Day 90
Meaningful agent conversations (cum.)	10–12	28–32	50–55
Scenario requests (cum.)	3–4	8–10	14–16
Applications (cum.)	1	3	5–6
Preapprovals (cum.)	0–1	1–2	3–4
A-tier agents	—	5–6	8–10
Closings (funded/CTC)	0	0	0–1 (informational)
SLA compliance	100%	100%	100%
Lab pieces live	0 (in review)	3	3 + cadence locked
Lab attributed conversations	0	0–1	0–2 (normal)

Sequence: Weeks 1–2: first-72-hours checklist (file 10, with contingency track), warm entries (6–8) into the sequence, half-dose wave. Week 3: full-dose gate (energy ≥6/10). Weeks 3–5: full 30 in motion; content clearing compliance. Weeks 5+: steady-state calendar (file 13). Day-60: F-lite go/no-go + IHDA answer resolved. Day-90: leading-indicator gate → scale or fix-one-thing.

Income reality across the view: ≈ $0. The first funded loan lands month 4 in base case. That is on plan, not failure — and it is exactly why the runway gate (§3.2) precedes everything.

7. Definition of success and failure

Day 30 — success: ≥3 scenario requests, ≥10 meaningful conversations, 100% SLA, approvals landed (or contingency track executed with escalations documented). Failure requiring action: <3 scenarios (revise the ask; formal redesign rule fires day 45), <7 conversations (diagnose energy vs. message first), any SLA miss pattern (freeze outreach, fix capacity), day-21 approval starvation (employer-viability review).
Day 60 — success: ≥8 scenarios cumulative, ≥3 apps, 5–6 A-tier agents, 3 Lab pieces live, IHDA/DPA answer in writing, F-lite decision made on evidence. Failure requiring action: conversations green but scenarios red 6 weeks (rework the ask — rule 3); walk-away trigger 2 tripping (§4).
Day 90 — success: ≥8 A-tier agents, ≥3 repeat scenario senders, ≥5 apps, SLA ≥95%, energy sustainable (L14 ≥6 avg) → scale (second wave, phase-2 geography prep). Failure requiring action: below gate → no expansion; fix the single worst funnel stage in month 4. Not failure: zero closings (month 4 is the base-case first funding), 0–2 Lab-attributed conversations, conservative-pace funded count with green leading indicators (that's a calibration update — rule 9).

Supersession note: files 00–24 and 27 were corrected in place during the revision pass and should agree with this file. If any residual conflict is found, this file governs and the other file gets fixed — log the fix in the day-30/60/90 gate notes.

PreviousExecutive Recommendation NextPrimary Business Plan