Mitchell Growth System← All documents

The Scenario Desk engine, the First-Home Lab, and the self-employed experiment in full detail.

09 — Primary Business Plan: The Scenario Desk + Southland First-Home Lab

Prepared 2026-07-22. Figures verified as of this date unless marked otherwise. Binding decisions: research_notes/DECISION_BRIEF.md. Model scoring: 08_strategy_model_scorecard.md. Everything Ready-specific is [UNVERIFIED — confirm with manager] until the manager meeting closes it out. Comp is always [BPS_COMP] — never invent a number.


1. Positioning: the honest newcomer

Positioning statement. Mitchell is a new loan originator in Tinley Park with a finance degree, a documented response-time standard, and the backing of Ready Mortgage Lenders' experienced operations team. He does not claim experience he doesn't have. He claims — and proves — preparation, speed, and transparency: hard scenarios answered in writing within four business hours, buyers prepared before they write offers, and no surprises at the closing table.

The honest-newcomer narrative, in Mitchell's own voice (approved framing):

"I'm new to originating — I won't pretend otherwise. What I bring is a finance degree from Illinois, an experienced lending team behind me, and more time per file than any 20-year veteran can give you. I'm not asking you to replace your preferred lender. Give me one difficult scenario or one buyer who needs a second look. I'll show you how I communicate and whether Ready has a useful path."

Hard rules (from DECISION_BRIEF, non-negotiable): - Never borrow Ready's institutional track record as personal experience ("we've closed thousands of loans" → prohibited in first person). - Never claim closing-speed numbers without operational verification and compliance approval. - Never repeat Ready's website puffery ("all loan programs in the market," "will get your loan closed!"). - Never position as an IHDA lender while Ready is not on the participating-lender list (verified NOT listed, list eff. 2025-12-18). - Every ad-like item carries: [MITCHELL_NMLS], Ready NMLS [READY_NMLS_1100518 — verify], www.nmlsconsumeraccess.org, Equal Housing statement/logo, company-required disclosures (38 Ill. Adm. Code 1050.940).

Geographic wedge: Tinley Park + Oak Forest + Orland Park + Lockport (phase 2 at day 90+: Frankfort–Mokena–New Lenox, then Homewood–Flossmoor–Matteson). Rationale in research_notes/local_market.md — attainable medians (Oak Forest $325k, Tinley $364k), 17–19 DOM velocity, the Harlem Ave agent cluster, Lockport's new-construction/tax-arbitrage leg, and Mitchell's own residency (Harmony Square, chamber, Boo Bash = free repeated visibility).


2. PRIMARY ENGINE — The Realtor Scenario Desk

2.1 Service definition

The Scenario Desk is a standing offer to a pilot list of ~30 buyer-active agents in the wedge: send any hard financing scenario; get an acknowledgment within 1 business hour and a written, guideline-checked one-page scenario brief within 4 business hours (published business hours: Mon–Fri 8:30–5:30 CT — the single public definition; the calendars' earlier internal start times are work time, not SLA time). Requests outside hours start the clock at the next open; Saturday requests get a same-day human acknowledgment and the brief by next business morning — Saturday is NOT in the published brief SLA until Ready's weekend coverage is verified in writing [UNVERIFIED — file 20 Q15; red-team C4]. The brief answers, in plain English: can this borrower likely qualify, under what product, what documentation will be needed, where the file is weak, what the realistic backup structure is, and what the buyer should resolve before writing an offer.

It is explicitly not a pre-approval, not a commitment, and not underwriting. It is structured analysis with the disclaimers to match. The desk's promise is a service standard, not an approval guarantee.

What qualifies as a scenario: self-employed income questions, gift/DPA stacking, condo warrantability doubts, DTI squeezed by Cook County tax escrows, credit-event seasoning, multi-unit house-hacks, appraisal-condition worries on 1950s–70s stock, "my lender went quiet" second looks. What gets escalated instead of briefed: anything needing an actual credit pull, rate quote, or borrower documents → moved into Ready's secure application process (§2.7).

2.2 Intake — fields and the no-sensitive-data rule

Intake by whatever channel the agent already uses (email/text/call), normalized by Mitchell onto a one-screen form (Google Form or CRM form — tool decision at [CRM_CHOICE], placeholder until stack file 12 is approved).

Fields: 1. Agent name / brokerage / best callback channel 2. Deadline context (offer tonight? listing appointment Thursday?) 3. Purchase price range + town 4. Property type (SFR / townhome / condo / 2–4 unit) + condo association name if applicable 5. Occupancy (primary / second / investment) 6. Approximate down payment available + source (savings / gift / DPA needed / sale proceeds) 7. Income type (W-2 / self-employed 12mo / self-employed 24mo+ / 1099 / retired / mixed) — ranges and types only 8. Approximate credit band (agent's words: "great / good / fair / bruised, event ~year") 9. Approximate monthly debts (ranges) 10. The actual question, in the agent's words 11. Anything already tried (prior lender's answer, if known)

The no-sensitive-data rule (printed on the form itself): "No SSNs, no DOBs, no account numbers, no paystubs, bank statements, or tax returns, no full borrower name tied to loan facts. Ranges and descriptions only. If a document is needed, we move to Ready's secure application link — never email or text." This is a GLBA/NPI firewall (see research_notes/ai_stack.md §8B) and it also keeps the desk fast: scenario analysis needs shapes, not documents. If sensitive data arrives anyway, Mitchell follows the AI-SOP incident steps (stop, delete, document, report same business day).

2.3 Response workflow

  1. T+0 to 1 business hour — human acknowledgment (not autoreply): "Got it — working on it. Brief to you by [time]. One question: [single best clarifying question, if any]." Sent by the channel the agent used. 1.5. Redaction gate — MANDATORY, before any AI window opens (red-team M7): Mitchell manually normalizes the raw agent message onto the intake template (§2.2) — initials/ranges/types only, no names, no addresses, no exact figures tied to an identifiable person. The normalized form, never the raw text/email, is the only thing AI ever sees. Real agents will text "John and Maria S., 8734 [street], he's 1099 at ~$95k, 612 score" — that message gets normalized by hand and the original is never pasted anywhere. This step is on the brief SOP checklist and is not skippable under SLA pressure; a missed SLA is recoverable, a GLBA incident is not.
  2. Analysis block (25–45 min): Mitchell works the normalized scenario against agency guidelines, the product matrix (research_notes/products_ihda.md §6), and Ready's confirmed IL menu [UNVERIFIED — confirm with manager]. AI use per the AI SOP: fictional-persona framing, public guideline research, draft structuring — AI is never the source of record; every number is verified against an authoritative source before it leaves draft.
  3. Review gate (v0/v1 — red-team C4): v0 (default, until a reviewer is confirmed): Mitchell's own structured self-review — checklist pass against guidelines, sources re-verified, weak points stated — and the SLA/scripts claim only "written, guideline-checked," never "reviewed with my team." v1 (after [MANAGER_OR_SENIOR_LO_NAME] is confirmed in writing as reviewer with a committed turnaround): every brief that names a product path is reviewed by them before sending; only then do the scripts' "human-reviewed by my team" lines (file 16 v1 variants) unlock. Review turnaround budget: 60 min [UNVERIFIED — confirm reviewer availability with manager; the claim is gated on this].
  4. T+≤4 business hours — brief delivered as PDF + a plain-text summary in the message body. The summary is terms-free (red-team M5.2): conclusion + next step + "full brief attached with details and disclosures" — no product terms, no down-payment percentages, no rates or payments (a "3.5% down" in forwardable plain text is a Reg Z trigger term detached from its disclosures). The PDF carries the full disclosure block; the summary must be safe to forward on its own.
  5. Same-day log: scenario logged (anonymized) in the tracker — town, product path, weak point, outcome. This log is the Lab's content mine (§4) and the KPI source (§2.9).

2.4 The one-page scenario brief format

Fixed template, one page, every time:

2.5 SLA language (public-facing)

"The Scenario Desk standard: I acknowledge every scenario within 1 business hour and deliver a written, guideline-checked brief within 4 business hours (Mon–Fri 8:30–5:30). Send something Saturday and I'll confirm I have it the same day, with the brief first thing the next business morning. If I miss a window, I tell you why and when — you will never chase me for a status. The brief is analysis, not an approval; when a file is worth pursuing, I'll move it into a secure application the same day."

(v1 upgrade — only after the reviewer and/or weekend coverage are confirmed in writing: "human-reviewed" may replace "guideline-checked," and Saturday hours may join the published SLA. Publishing capabilities before they exist is the "broken promise worse than no promise" trap — red-team C4.)

No guarantee words ("guaranteed," "always," "fastest") — the standard is stated as a practice with a miss-handling rule, which is both honest and compliant.

2.6 The trial ask script (pilot outreach)

"Hi [Agent], this is Mitchell [LASTNAME] with Ready Mortgage Lenders — I'm a loan originator here in Tinley Park, and I'm calling to ask for something small. [Illinois opener satisfied: name + company + purpose, 815 ILCS 413/15.] I'm not asking you to replace your preferred lender. I run what I call a Scenario Desk: you send me one difficult scenario — the self-employed buyer, the condo that got weird, the file another lender went quiet on — and I get you a written answer the same day: what works, what's needed, where it's weak, and a backup structure. I emailed you an example brief [Monday] so you can see exactly what you'd get — the product demos itself. No cost, no obligation, no pressure on the buyer. If the brief is useful, use me again; if not, you've lost one email. Is there a deal on your desk right now that fits that?"

If yes → intake questions live on the call. If no → "Fair enough — the example brief in your inbox is yours either way; if a weird file ever lands, that's what the answer looks like." If no interest at all → thank, log, suppress from calls, keep on the quarterly useful-asset list unless opted out entirely.

Sample-brief-first rule (red-team M3): the Day-1 email leads with a finished synthetic sample brief attached (2–3 polished synthetic briefs exist by Day 4 — file 10 item 8). The brief is the proof; the call references it instead of promising it. "Can I send you an example?" is no longer the fallback — it is the opening move.

2.7 Escalation to Ready

The moment a scenario becomes a real buyer: "The next step is a short secure application so this stops being hypothetical — here's the link." Application via Ready's secure link only; no documents by email/text; credit pulled only with borrower authorization inside company systems. The brief's backup-structure section becomes the loan-structuring conversation. Handoff target: same business day as agent's go-ahead.

2.8 Follow-up cadence

2.9 The Scenario Desk plays — full 12-element specification

Play D1 — Pilot-list build and first-touch sequence 1. Who: ~30 buyer-active agents/small teams (25–40 band). Entry order is warm-first (red-team M3): the first 6–8 slots are Crosstown Hybrid/non-Flex agents reached via the family connection, plus family-network introductions to agents at other brokerages (conflict hygiene per file 15 §1: no Flex-team members, disclosure of the family relationship where relevant, no lender-steering entanglement). The cold corridor follows: the Harlem Ave cluster — RE/MAX 10 (~300 agents), KW Preferred (MC 644), CB Southwest (100+ associates), RE/MAX Synergy — plus Baird & Warner Orland and 4–6 Lockport/Will-side agents (Three Rivers AOR territory). Scored on public data: buyer-side closings, price band $250–450k, condo/attached activity, responsiveness signals. 2. Why attractive: densest agent concentration in the footprint within ~2 miles; buyer-side agents in the attainable band hit financing problems weekly; small teams lack an in-house lender relationship the way mega-teams have. 3. Value provided: a free, fast, written second opinion that makes the agent look prepared to their client. 4. Exact action: build the scored list (wk 1), then run the 14-day sequence per agent, 5 new agents entering per week, calls in one 60–90 min sprint per day max. 5. Sample message: script in §2.6; email version opens with one specific true observation about the agent's recent listing mix — no fake familiarity. 6. Channel + cadence: email day 1 → manual call day 3 → ≤20-sec voicemail same day if no answer → LinkedIn day 6 → useful asset day 10 → in-person (office preview, open house visit) day 14 where natural. 7. Time: wk 1: 6 hrs list-build; then ~7 hrs/wk (5 sprints + prep). 8. Tool: Claude for prospect research + list scoring; [CRM_CHOICE] for sequence tracking; phone is manual, human-dialed only. 9. Deliverable: scored pilot list + per-agent touch log. 10. KPI: meaningful conversations/wk (target 5 by wk 3); scenario requests (north star — target 4 in first 30 days). 11. Stop/continue: after 2 full weeks of completed sequences, if meaningful-conversation rate < 10% of dials, stop and rewrite the script with role-play (ChatGPT Advanced Voice) before resuming; do not push volume through a failing script. 12. Compliance check: manual live calls 8am–9pm only; Illinois opener (name + company + purpose); scrub the federal DNC registry anyway, treat agent cells as potentially residential; no dialers/RVM/blast texts; CAN-SPAM footer + NMLS block on any templated email; suppression list honored across channels (per compliance_outreach.md one-page summary — the whole page governs this play).

Play D2 — The Scenario Desk service itself 1. Who: the pilot agents who bite, plus inbound referrals from them. 2. Why: hard scenarios are where preferred-lender loyalty is weakest — the incumbent already fumbled or went quiet. 3. Value: §2.4 brief on the §2.5 SLA. 4. Action: run workflow §2.3 for every scenario, no exceptions to the SLA; when a window will be missed, notify before it lapses. 5. Sample message: acknowledgment and brief formats in §2.3–2.4. 6. Channel + cadence: agent's channel; SLA cadence; §2.8 follow-up. 7. Time: budget 90 min/scenario all-in; capacity cap 3/day in month 1 (protects SLA and review gate). 8. Tool: Claude (guideline research, draft structure — SOP rules), product matrix, template; reviewer gate. 9. Deliverable: one-page brief PDF + anonymized log entry. 10. KPI: briefs delivered; SLA hit-rate (target ≥95%); brief→application conversion (target ≥20% by day 60); repeat-request rate per agent (the true trust metric). 11. Stop/continue: at day 45, if <6 total scenario requests despite ≥25 completed outreach sequences, the offer isn't landing — pause D1, interview 5 friendly agents on why, redesign the offer before spending more outreach hours. Do NOT quietly widen to generic "send me your buyers" begging. 12. Compliance check: disclaimer block on every brief; no rate quotes or payment figures in briefs unless from company-approved materials with approved assumptions; NPI firewall per §2.2; RESPA — the desk is Mitchell's own competence given freely to the marketplace, never conditioned on referral volume, never paired with gifts or paid agent expenses.

Play D3 — Preparation-before-every-call ritual (personality design) 1. Who: Mitchell himself, before every outreach sprint and agent meeting. 2. Why: the brief's personality rules — prep converts social load into analytical work, his strength. 3. Value: call quality that reads as senior. 4. Action: 15-min pre-sprint block: review each agent's log, one specific opener each, one likely objection each; 10-min post-sprint debrief from written notes (no unapproved recordings). 5. Sample: objection drill — "We have a lender we love" → "Keep them — that's not the ask. The ask is one scenario your lender found awkward. If my brief isn't the most useful thing in your inbox that day, we're done." 6. Channel + cadence: ChatGPT Advanced Voice role-play daily, 15 min, against approved scripts only. 7. Time: ~2.5 hrs/wk inside the D1 budget. 8. Tool: ChatGPT Plus Project (approved materials only), written debrief log. 9. Deliverable: evolving objection playbook v-dated per the file-naming SOP. 10. KPI: objection-conversion notes; subjective call-confidence 1–5 logged weekly. 11. Stop/continue: if role-play stops changing call outcomes by wk 6, cut to 2×/wk and reallocate to brief quality. 12. Compliance check: no real borrower or agent personal data in role-play; fictional personas only (AI SOP §A3).


3. SECONDARY ENGINE — Southland First-Home Lab

(Working name — requires Ready naming/branding approval before any public use. Fallback: publish under "Mitchell [LASTNAME], Ready Mortgage Lenders" with no brand name.)

A local education brand for first-time buyers in the wedge, built on three content assets nobody in the footprint publishes (verified gap, local_market.md §10 — SERPs are branch pages and directories; the only individual-MLO content found is 2017-era).

3.1 Asset 1 — The DPA Stack Guide ("Down-payment help in Tinley Park, Oak Forest, Orland Park & Lockport — what actually exists in 2026")

Town-specific, plain-English guide to layering: IHDA Access (Home 6%/$15k deferred; Forgivable 4%/$6k; Deferred 5%/$7.5k; Repayable 10%/$10k — 640 min score, $1,000-or-1% borrower contribution; 2026 income/price limits must be pulled live from ihdamortgage.org/limits before borrower use — the findable $134,520/$610,939 figures are 2024-vintage), Cook County DPA (reopened 2026-07-20 — up to 5%/$25,000 forgivable, 5-yr, income <120% AMI, Cook-side buyers only — Lockport buyers are NOT eligible, say so plainly), and FHLB Chicago Downpayment Plus ($10k grant, ≤80% AMI, via member banks). Honesty constraint, stated in the asset itself: Ready is not currently an IHDA participating lender; the guide teaches what exists and who offers it, and explains exactly what Mitchell can and cannot do until that changes [UNVERIFIED path — manager question #1]. Counterintuitively, naming competitors who hold IHDA approval is the credibility play — it proves the guide serves the reader.

3.2 Asset 2 — Cook-vs-Will tax math ("Your 2026 reassessment survival guide: what a Tinley Park payment really costs vs. Lockport")

2026 is the south/west-triad reassessment year; the 2023 precedent produced a record 19.9% median south-suburb bill jump. Will-side effective rates run ~2.3–2.8% vs 3%+ in much of south Cook — on a $360k Tinley home the escrow line approaches half the P&I ($300k loan @ 6.55% PMMS 2026-07-16 ≈ $1,906/mo P&I; taxes ≈ $900/mo). The asset shows side-by-side full-payment math for the same buyer in Tinley vs Lockport vs New Lenox, with every assumption printed and dated. This is the honest-broker conversation no local lender publishes, and it feeds the Lockport/new-construction leg (Lennar Oak Valley townhomes from ~$381k). Rate/payment figures in a published asset = advertising: company-approved assumptions and compliance sign-off required before publication.

3.3 Asset 3 — Condo triage ("The condo financing rules just changed. Here's the new playbook for Tinley and Orland.")

~1 in 5 units of Tinley/Orland stock is attached. 2026-08-03 — days from plan start, and almost certainly past by the time this asset clears compliance (red-team M6: the original "publish before August 3" framing was arithmetically dead — naming approval + disclosure block + compliance review + writing cannot fit in the gap). The asset is deliberately framed post-change, which is the stronger hook anyway: Fannie/Freddie eliminated Limited Review (used in ~40% of condo reviews) — every condo loan now needs full review or waiver, more HOA documentation, est. 2–4 weeks added — and most agents haven't felt it yet. "Here's what just changed on August 3 and what it does to your attached listings now" opens more doors than a countdown ever did, and it stays true for months. The forward deadline is real and unexpired: 2027-01, when the minimum reserve line rises to 15%, pushing thin-reserve associations non-warrantable — that is the urgency line the asset may use. Content: a buyer/agent checklist for pre-screening an association (reserves, special assessments, litigation, owner-occupancy) before going under contract, and what to do when a project is on the non-public unavailable list. Still the Scenario Desk's best door-opener with listing agents; still publishes first. No script or asset anywhere in this plan says "act before August 3."

3.4 Lab plays — full 12-element specification

Play E1 — The three cornerstone guides 1. Who: first-time and move-up buyers renting or owning in the wedge (largest renter pools: Lockport 15.4%, Tinley 13.1%, Oak Forest ~17% [flagged estimate]); secondarily the pilot agents, who get every asset first. 2. Why: verified zero-competition topics with built-in 2026 urgency. 3. Value: the only town-specific, dated, source-linked answers to the three questions local buyers actually ask. 4. Action: publish in order — condo triage first (wk 1–2 of content production, post-change framing per §3.3), DPA stack (wk 3–4), tax math (wk 5–6, after approved payment assumptions exist). 5. Sample (condo-asset opener, post-change): "If you're buying a condo in Tinley Park or Orland Park, the federal review rules changed on August 3, 2026 — and your association's paperwork, not your credit, may now decide whether your loan closes. Most buyers will find out at underwriting. Here's the 10-minute check to run before you write an offer — and the second deadline (January 2027) that will catch thin-reserve associations next." 6. Channel + cadence: simple site page per asset ([SITE_STACK] placeholder) + PDF version for agents + a 60–90-sec vertical video summary each; one asset shipped every 2 weeks, then monthly refresh of dates/figures. 7. Time: 4–6 hrs/asset writing + 2 hrs video + compliance turnaround; batched per the personality rules. 8. Tool: Claude (research/drafts, SOP rules), Canva or equivalent, phone camera. 9. Deliverable: 3 live pages + 3 PDFs + 3 short videos, all carrying the full NMLS/Equal Housing block. 10. KPI: scenario requests and workshop signups attributed to an asset (ask "how'd you find me?" every time); page → contact conversions; NOT impressions. 11. Stop/continue: at day 90, if the three assets have produced zero attributed conversations, do not write more assets — redistribute the existing three harder through agents and chamber channels for 30 days before deciding. Honest expectation set now (red-team M8): 0–2 attributed conversations by day 90 is NORMAL for quarter-one content, not failure. The Lab is a trust-and-referral asset with a small, slow direct yield in year one — it is not near-term pipeline and is never sold internally as such. If the DPA lane fails verification at the manager meeting, the Lab's segment priority recuts per file 26 §DPA-branch. 12. Compliance check: every asset = an advertisement (NMLS IDs + nmlsconsumeraccess.org + Equal Housing; Reg N archive ≥24 months — archive 5 yrs); no self-produced rate/payment content without approved assumptions; no IHDA-lender implication; fair lending — assets serve the whole footprint, no geographic/protected-class proxy targeting in any paid boost (special ad category if ever boosted).

Play E2 — First-Home Lab workshop (small-room format) 1. Who: 8–15 renters/first-timers per session, recruited via the assets, chamber, library, and pilot agents (agents may invite clients; agents' costs are never paid — RESPA). 2. Why: small-room teaching is the highest-trust, lowest-social-drain conversion format for Mitchell's temperament; the DPA-stack material is a natural 45-min curriculum. 3. Value: a genuinely useful, no-pitch hour: "what a payment really costs here, what down-payment help exists, what to fix before you apply." 4. Action: monthly, starting month 2 — after the DPA asset is approved; venue: Tinley library meeting room or chamber space at posted public rates, paid by Mitchell/Ready per policy [UNVERIFIED — confirm marketing budget with manager]. 5. Sample invite: "Free first-home workshop, Tinley Park — 45 minutes on what a home here actually costs each month, the down-payment programs that exist in 2026, and the 5 things to fix before you apply. No sales pitch; bring questions. Seats limited to 15." 6. Channel + cadence: asset pages, chamber newsletter, library posting, agent forwards; monthly. 7. Time: 6 hrs/month (prep 2, deliver 1.5, follow-up 2.5) + recovery block scheduled after, per personality rules. 8. Tool: slide kit from the assets; paper signup sheet with explicit opt-in checkboxes for follow-up. 9. Deliverable: repeatable 45-min curriculum + attendee follow-up sequence (opt-in only). 10. KPI: attendees; opt-in rate; consultations booked within 14 days (target 2/session by session 3). 11. Stop/continue: if session 3 still books zero consultations, stop the public format and convert to agent-hosted client sessions only (agent supplies audience, Mitchell supplies teaching — with compliance review of the arrangement first). 12. Compliance check: workshop materials are ads (full disclosure block); no borrower NPI collected in the room beyond opt-in contact info; express written consent captured if any text follow-up; venue costs never routed through an agent; any co-hosted event needs prior written compliance approval with FMV documentation.

3.5 Lead capture rules

Opt-in only, value-first: each asset offers one upgrade (e.g., "the condo pre-screen checklist as a fillable PDF") for an email address, with plain language about what they'll receive and a working unsubscribe honored same-day (legal max 10 business days; Mitchell's standard is same-day). No purchased lists, no scraping, no retargeting pixels without compliance review. Contacts enter a monthly useful-asset email (full CAN-SPAM footer + NMLS block), never a daily drip. Anyone asking a loan-shaped question gets a human reply and, when appropriate, the secure application link — the same NPI firewall as the desk: nothing sensitive by email.


4. The F-lite experiment — Investor/Self-Employed File Lab

Definition (dosage-controlled): one content franchise plus reps, not a pivot. - 2–3 self-employed/investor scenario drills per week run through the Scenario Desk template as practice files (fictional or anonymized), building bank-statement/DSCR fluency before real ones arrive. ~2 hrs/wk. - One "File Lab" content piece every 2 weeks (anonymized/synthetic): "Contractor with great deposits and an ugly Schedule C — can he buy in Lockport?" Published under the Lab with the same ad-compliance block. ~2 hrs per piece. - Distribution: Tinley chamber (400+ members) + the multi-chamber network + pilot agents who work with trades buyers. No paid ads, no dedicated brand, no outbound campaign. - Hard precondition: the first real bank-statement file requires Ready's confirmed IL non-QM menu and investor list [UNVERIFIED — manager question #2]. Until confirmed, F-lite is training + content only, and every piece says "programs vary by lender; availability confirmed at application."

Day-60 evaluation (binary, scheduled now for [WEEK_1_START_DATE] + 60 days, adjusted to the nearest business day — no hardcoded date per the file-12 rule; red-team m2): - Scale it if ≥2 real self-employed/investor scenario requests arrived through the desk or the content, AND Ready's non-QM menu is confirmed in writing → F-lite graduates to a named third content pillar with 4 hrs/wk. - Kill it if <2 real requests, or Ready cannot support the products → archive the content (it still serves SEO), stop the drills, fold the hours back into D and E. No zombie middle state.


5. How the engines feed each other

The plan is one machine, not three projects:

  1. Desk → Lab: every anonymized scenario log entry is a pre-validated content topic — a real question a real local agent asked. The Lab never guesses what to write; it publishes what the desk already answered. (Master-prompt franchise fit: "Can This Deal Close?" = desk logs; "Realtor Rescue Minute" = desk weak-point sections.)
  2. Lab → Desk: every asset ends with the same door: "Working with an agent? Have them send me the scenario — written answer in 4 business hours." Assets give pilot agents a reason to remember the desk between deals, and give Mitchell a non-needy touch for follow-up cadence (§2.8).
  3. Lab → applications directly: workshop attendees and guide readers without an agent become buyer consultations — and Mitchell introduces them to pilot agents, reversing the referral flow. (Introductions are made on fit, never traded against loan referrals — RESPA: no quid pro quo, and no tracking of "balance owed.")
  4. F-lite → both: trades-corridor content differentiates the desk with investor-friendly agents; desk drills make the content credible.
  5. One calendar: desk sprints and briefs in morning analytical blocks; content batched in two afternoon blocks/wk; workshop monthly; recovery blocks after high-social days. Baseline 45–50 hrs/wk with the 35–40 hr alternative mapped in the calendar file (file 13, forthcoming). Sundays off or 30-min planning only.

6. Why this holds up for 12–24 months


7. What Mitchell will NOT do — and why

  1. No Zillow Flex. ~$2,600 gross per $350k closing before costs, evenings-and-weekends speed-to-lead duty, a "Zillow Home Loans contact rate" KPI, and a live steering class action — structurally misaligned with being Ready's MLO. If the broker license should stay active: Crosstown Hybrid 75/25 (no Flex leads) or Fathom/eXp as parking, only with Ready's written outside-business-activity approval.
  2. No active dual-track production, and never both roles on one transaction in year one. USDA prohibits it; FHA/VA/investor overlays and the Reg Z comp question are unresolved. The broker license is literacy and optionality, not a second job.
  3. No paid leads, no mass outreach platforms, no dialers/RVM/AI-voice/blast texts. The compliance page bans most of it outright (TCPA has no B2B exemption for automation; FCC 24-17 makes AI voice "artificial"), and the desk's economics beat purchased leads anyway.
  4. No MSAs, no co-marketing spend, no agent gift patterns, no paying agent expenses. RESPA §8 has no de minimis exception; the desk gives time and competence, never money-equivalents. Any specific co-marketing idea goes to compliance in writing first.
  5. No reverse, foreign-national, construction, hard-money, or commercial origination. Defer/refer — high complexity or high optics risk for a rookie; content may mention them honestly ("I refer these") but Mitchell does not originate them yet.
  6. No IHDA-lender positioning until Ready is on the list. Teach the programs, name who offers them, push the approval question internally.
  7. No borrowed experience, ever — no company track record in first person, no closing-speed claims without verification, no fake familiarity in outreach, no Ready-website puffery repeated.
  8. No UIUC alumni program before month 3, and then only as a designed experiment with its own stop rule — alumni affinity alone is not demand.
  9. No Kimi in the workflow (China-hosted, trains on inputs, compelled-disclosure exposure; expect the company "no" — the stack loses nothing without it). Core stack stays Claude Pro + ChatGPT Plus, $40/mo, under the written AI SOP — which itself becomes a differentiator as Fannie LL-2026-04 pushes seller AI-governance expectations.

Every "not now" has a re-entry condition (day-90 review for geography phase 2 and UIUC; manager confirmations for products; written approvals for license parking). Nothing is banned forever; everything is sequenced on purpose.

PreviousRevised Final Plan (Plan of Record)NextStrategy Model Scorecard