Top Opportunities — Final Ranked List

Sprint: Sun Oct 25, 2026, ~14:11–16:15 America/Bogota · Waves W1–W7 + parallel build-once app sprint (W1–W6)
Rule Carlos set (Oct 25, 15:37 COT): not capped at 10. Every candidate that genuinely clears the bar is in; no padding.

TN flag — how to read it (Carlos’s rule, Oct 25 16:12). The flag is information only: flagged ideas stay in the list, ranked on merit, never removed, frozen or demoted for it. W6 upgraded this from a flag to a hard gate: USCIS Jun–Jul 2025 guidance bars TN self-employment including sole/controlling ownership, and deposits are revenue. A Mom-CCPC does not cure de facto control. Counsel scoped to the validation tests themselves (who solicits, who performs, which entity invoices, whose account receives) must come first, in writing, per test — before any deposit, paid pilot or paid audit is taken. Every personal-revenue entry below is flagged where this applies. If counsel blocks Carlos's personal participation across the board, the portfolio contracts to the routes counsel did clear (Mom-owned, Maheva-owned, capital-only) — it does not proceed on deposits while "the structure is being figured out."

Why 18. W7's corrected portfolio proposed 15 (core 10 + also-cleared 5). After W7 was drafted, the parallel app sprint finished with 4 validated finalists that had been promised inclusion. Two of W7's weakest also-cleared entries are therefore demoted to the conditional watch list below (Ontario AODA — clears on deadline but loses its operator, Mom's third business; ChatGPT Ads desk — clears only on an unverified binary billing-entity gate), and the 4 app finalists are added: 15 − 2 + 4 = 17 counted (core 12 + also-cleared 5). Nothing W6 cut was resurrected to reach this number.

Labels: [est] = estimate / planning figure. Poster claims are labelled as such. Monthly potential is the W7 base case at 6 / 12 months unless noted. All forward figures are estimates, not measurements.

At a glance

#	Candidate	Who runs it	Cost to start	First money	Monthly potential @6–12 mo [est]	Carlos hrs/wk
1	Structured income sleeve (repaired)	Capital only	$50k allocation (of $100k envelope)	Immediate	~$160–215 reliable on $50k tranche	~0
2	Primera Cuenta — repaired (Latino SMB bookkeeping)	Mom	CAD ~1.5–3k all-in	2–4 wks	CAD 2.5–3.5k base, cap-bound	1–2
3	Ficha Perfecta — repaired (setup sprint + 90-day Care)	Maheva + agents, Carlos QA	~$100–350	3–7 days	~$2,700 base → $3–4k	0.5–1
4	GEO/AEO finite fix sprints	Carlos + Maheva (FR book)	~$330–900	1–3 wks	$3–5k base → $5–7.5k with white-label	2–3
5	Cabinet Augmenté seasonal bilan pack (capped)	Maheva	Counsel €300–600	Nov window only	Seasonal €2.5–5k total, not MRR	~1
6	Skincare private label — 1 SKU first	Maheva + agents	$3–6k test run (gated $12k tranche)	6–10 wks	$2–4k base → $5–8k	~2
7	Raíces Desk (diaspora family verification)	Mom lead, Carlos verifies	$600–1,000	2–4 wks	$3–4.5k base → $3.5–6k	2–3
8	Colombian coffee micro-lots → FR roasters (pre-sold only)	Maheva sales, Carlos origin	$2–4k, only vs pre-sale deposit	6–10 wks	$1–3k averaged, lumpy — not MRR	~1
9	Medellín mid-term apartment (conditional)	Capital + local manager	$2k diligence; purchase ≈$148k separate	6–12 wks post-close	~$380–435/mo stressed [est]	1–2
10	Expat Auto-Entrepreneur Desk	Maheva	€200–400	1–2 wks	€1.5–2.5k base → €2–4.5k	1
11	RutinaGrid (app — PRIMARY first build)	Carlos + agents, AI loop maintains	$200–700 build	Late Nov / Jan-1 window	~$937 net base @M6 [est]	~0.5
12	Cartas FR↔ES (app — first to test)	Maheva QA + agents build	≤$200 pre-sale; $250–750 build	Pre-sale, wks 2–3	~$2,070 net base @M6, conditional [est]	~0.5
13	Operator, Not Coder ⚑ TN-flagged	Carlos	~$0–500	1–3 wks	$5–10k first plateau [est]	4–6
14	DocuES (app — Alternate 2)	Carlos + agents, AI loop	$350–1,050 build	~6–8 wks	~$2,271 net base [est, soft input]	~0.5
15	PruebaSocial (app — gated #4)	Carlos + Maheva onboarding	$250–450 build	Gated	~$972 MRR @M12 base [est]	0.5–0.75
16	Southbound Setup Desk	Maheva ops, Carlos design	~$800–1,100	2–3 wks	$3–6k base, lumpy/seasonal	3–5
17	Arrière-Boutique trilingual pods (as a paid test)	Maheva QA, Mom CCPC invoices	CAD 1–2k	3–6 wks	CAD 3.6–5.4k base @6 mo	2
18	Borderline — repaired, firm-channel only	Carlos	$500–1,500 validation	6–12 wks	$1–3k lumpy consultancy [est, weakest number]	3–5

Core list (12)

1. Structured income sleeve — REPAIRED, relabelled as the control

What: 50%+ T-bills/SGOV (4.22–4.44%, Oct 2026) + JEPI (7.21% distribution) / JEPQ (11.64% distribution, capped ≤25%) blend — cash management with daily liquidity, not a business.

Who: Capital only · Carlos ~0 hrs/wk (quarterly check).

Cost: $50,000 allocation inside the $100k envelope (not sunk cost; the full blend was modelled on $100k). First money: Immediate on execution; first distributions within ~1 month.

Monthly potential [est]: On the $50k tranche ~$225–285/mo gross-to-net, plan on ~$160–215/mo reliable after withholding + distribution volatility. (On $100k: ~$450–570 gross-to-net, ~$320–430 reliable — the old $600–1,000/mo line was killed in W6.) Only the T-bill leg (~$180/mo per $100k) is true risk-free interest.

First validation test: No sales test. One written accountant answer first: JEPI/JEPQ withholding in Carlos's actual filing position — Canada treaty 15% vs Colombia no-treaty 30% is unresolved, and a 15-point swing beats any fund choice. No allocation before that answer.

#1 kill risk: Distributions ≠ return — NAV give-back on the covered-call legs can hide inside "yield." Judge quarterly on total return vs SGOV-only; 2 consecutive quarters of underperformance = rebalance trigger. Gate 0: clean (capital only).

2. Primera Cuenta — REPAIRED to established Latino SMB bookkeeping (Mom-run)

What: Spanish-first bookkeeping for already-established GTA Latino SMBs at CAD 399–599/mo (market band CAD 400–800/mo, W5B pricing guides). The CAD 349–499 newcomer pack is demoted to a lead magnet — free CVITP/settlement agencies own that segment (W6).

Who: Mom (owner-operator, Canada — cleanest Gate 0 in the sprint) · Carlos 1–2 hrs/wk (systems/funnel only) · Mom 8–12 hrs/wk, capped at 5 founding clients until a part-time Spanish bookkeeper is hired.

Cost: Validation CAD 500–1,000 [est]; $3,000 setup tranche. First money: 2–4 weeks [est]; deposits possible in the test week.

Monthly potential [est]: Low CAD 1,200–2,000 (3 clients) / base CAD 2,500–3,500 at the 5-client cap; CAD 4,500–7,500 at 12 mo only if the part-time bookkeeper is hired (W5B's CAD 4–9k is a ceiling, not a base).

First validation test: Pre-conditions first (Week 0, ~$0): Mom's Ontario designation, EFILE eligibility and professional insurance verified, plus a written scope sheet (bookkeeping + admin/referral only — no immigration or benefits advice). Then: PASS = ≥3 paid CAD 100 deposits in 7 days for bookkeeping (newcomer-pack deposits alone do not pass).

#1 kill risk: Mom's capacity — she is 65, still working a hard job, and the Jan–Apr T1 surge hits the same person. The 5-client cap is the business model, not a growth failure.

3. Ficha Perfecta — REPAIRED (setup sprint + Care as a 90-day block)

What: Done-for-you Google Business Profile fix sprint for Medellín SMBs (COP 590k / ~$185, 72h) + ongoing Care pre-sold as a 90-day paid block (COP 190k/mo × 3 upfront), never an open-ended review promise.

Who: Maheva sends/judges, agents audit + deliver, Carlos QA/payments · Carlos 0.5–1 hr/wk · Maheva 5–8 hrs/wk.

Cost: Validation $0–10 / launch $100–300. First money: 3–7 days — the fastest cash bridge in the portfolio.

Monthly potential [est]: ~$2,700/mo gross base at 6 mo (8 setups + 20 Care clients) → $3,000–4,000 base at 12 mo. The unproven $1,499 multi-location "Reputation Sprint" bundle is upside only. Best revenue per Carlos-hour of any business here.

First validation test (sales test #2 of the permitted two): 30 screenshot-led WhatsApps to Medellín businesses with a visible profile defect. PASS = ≥2 setups PAID by day 7. Kill if <6 replies total — do not re-run the same list cheaper.

#1 kill risk: Google's Apr 2026 profile gating/suspension surge (W6) — the reason Care is a prepaid 90-day block. Gate 0 applies — counsel-cleared route before deposits.

4. GEO/AEO — REPAIRED to finite fix sprints

What: ES/FR AI-visibility fix sprint at $1,499–2,499, delivered in ~7 days, plus an agency white-label variant (agency retails ~€2,499, delivered wholesale at $500–750/end client). The $199–365/mo score-watch retainer is dead (non-deterministic measurement, $29–99 self-serve tools — W6); a $199 audit survives only as a credited entry/downsell. The category is the breakout Stripe-verified one from W1B (AEO Engine $156k MRR; RankAI $56k MRR from 155 subs — TrustMRR, Oct 2026), but Profound's $180M Series D (Sep 2026) saturates English, so the raise path is LatAm/EU, not SV.

Who: Carlos (ES partner close) + Maheva (FR book), agents deliver · Carlos 2–3 hrs/wk · Maheva 3–5 hrs/wk.

Cost: Validation $29–99 / launch $300–800. No SaaS build before 5 paying clients. First money: 1–3 weeks (paid sprint, not a free report).

Monthly potential [est]: Base $3,000–5,000 at 6 mo → $5,000–7,500 at 12 mo only if the white-label channel or a second delivery agent exists. Direct + white-label share one delivery pool — never sum them.

First validation test (slips to Nov 2–3 under the max-2-tests rule): 25 screenshot-led targets + 30 agencies. PASS = ≥3 paid $199 audits credited toward sprints AND ≥1 sprint sold at ≥$1,499 within 14 days, OR ≥2 agencies agree to offer it to their book within 10 days. Fail = replies only at the $199 tier → GEO folds into the other services as an add-on.

#1 kill risk: Buyers engage only at the $199 tier, and the measurement objection (AI answers are non-deterministic) undermines the promise — finite sprints with reproducible before/after evidence are the repair, and the test proves or kills them.

5. Cabinet Augmenté — seasonal bilan pack, KEEP only capped (runs FIRST in calendar time)

What: Seasonal bilan pack for French CGP cabinets (€490 / 10 bilans, €990 / 25, €59 à la carte; setup credited toward a €299–499/mo subscription before Jan 31), max 5 cabinets, signature deadline Nov 20.

Who: Maheva only (her wealth-adviser credibility is the pitch) · Carlos ~1 hr/wk · Maheva ~8 hrs/wk in the window.

Cost: Validation = counsel €300–600 [est] / launch €0–300. First money: Deposit possible Oct 31–Nov 13 if counsel clears; otherwise $0 this season.

Monthly potential [est]: Not MRR — a season total of ~€2,450–4,950 for 5 packs; at 12 mo, €1,500–3,000/mo base only if 5–8 subscriptions survive past Jan 31. Probability-weight every planning number at ~40–50% for the counsel gate.

First validation test (sales test #1 of the permitted two): Counsel brief sent Mon Oct 26. Interim pass (Oct 31) = counsel engaged with written sign-off date ≤Nov 6 AND ≥5 sales calls booked from ≥20 warm messages. FINAL PASS (Nov 13) = ≥2 packs PAID (50% deposit). Auto-kill: no counsel-cleared invoicing route by Nov 6 — re-aim at the Apr–Jun déclaration season.

#1 kill risk: Counsel is a single point of total failure against a hard Nov 20 window. Rank 5 here, first in sequence — sequencing ≠ ranking.

6. Skincare private label — REPAIRED to 1 SKU first (the Longevity Lab's pulled-forward validation)

What: ONE SKU — ceramide moisturiser, least claim-risk — made in Colombia (Tecnolabs / San Roque maquilas; MOQ 250, INVIMA notification ~1 day per prior supplier research). GHK-Cu only after a claims review; the other two hero SKUs are deferred. This is the AI Longevity Lab's product-first Phase 0 evidence, not a standalone clinic bet.

Who: Maheva (clinic/influencer outreach) + agents · Carlos ~2 hrs/wk · Maheva 8–10 hrs/wk in launch weeks, ~5 steady.

Cost: Validation $50–150, $0 production spend in the test / launch $3,000–6,000 [est] for a 250-unit test run — if the margin only lives at 1,000 units, the honest number is ~$8,000–12,000; get 250-vs-1,000 pricing in writing first. $12,000 gated tranche. First money: First real sale 6–10 weeks [est].

Monthly potential [est]: Base $2,000–4,000 at 6 mo → $5,000–8,000 at 12 mo with a second SKU + derm co-brand added only after the 1-SKU pass. The old 3-SKU $10–20k/mo line required US export and is not a base case.

First validation test: In-person clinic round with hardened LOIs. PASS = LOIs totalling ≥300 units with clinic margin % AND payment days written in (real terms may be 40–50% off retail + 60–90-day payment / consignment) plus maquila answers in writing: NSO holder = brand entity, formula ownership/transferability, Resolución 0814/2026 no-animal-testing declaration capability.

#1 kill risk: Clinic distribution — 0 committing clinics, or "yes" only on consignment terms that collapse the margin, kills the channel. Judged on units + terms in writing, not yes-count.

7. Raíces Desk — Mom-led diaspora family verification

What: Spanish-first verification + coordination + reporting for Colombians in Canada/US whose parents, property and paperwork are still in Colombia: property/parent checks with photos, health-panel days, notary/pension document shepherding via partner professionals. Single case $249–600; family plan $199/mo. No money transmission, no property management, no medical decisions. (Remittances to Colombia hit a record US$13.098B in 2025 — Banco de la República via W5A.)

Who: Mom lead in Canada · Carlos 2–3 hrs/wk (Medellín verification days) · Maheva/agents 2–3 hrs/wk. Best Carlos-leverage in W5.

Cost: $600–1,000 [est]. First money: 2–4 weeks [est].

Monthly potential [est]: Base $3,000–4,500 at 6 mo → base $3,500–6,000 at 12 mo. W5A's $6,000–11,000 is a ceiling, not a base.

First validation test: "Medellín verification week" offered to 15 diaspora contacts / GTA Colombian community groups — property or parent check, photo + written report in 7 days, $249. PASS = ≥3 paid cases in 10 days. Fail = warm interest but "a relative does it free."

#1 kill risk: Scope explosion into elder-care, legal and property-management liability — held in check only by the verification/coordination/reporting boundary in writing. Operator note: Raíces is Mom's second business here (with Primera, and AODA on watch) — sequence Mom's, never stack them.

8. Colombian coffee micro-lots → French roasters — pre-sold only

What: Asset-light B2B bridge: traceable single-farm micro-lots sourced in Colombia, pre-sold in French to independent roasters/épiceries, $1.50–3.00/kg bridge margin on 300–700 kg lots [est]. No inventory without a pre-sale deposit; exporter-of-record partner is non-optional. (EU coffee imports 2025: 2.9M tonnes / €18.7B, Colombia 6% — Eurostat via W5B.)

Who: Maheva (FR sales) · Carlos ~1 hr/wk averaged · agents (lot sheets, EUDR dossiers).

Cost: Test $300–600 / launch $2,000–4,000, first-lot freight only against a pre-sale deposit. First money: 6–10 weeks [est].

Monthly potential [est]: $1,000–3,000/mo averaged at 6 mo → base $3,000–5,000 averaged at 12 mo (W5B ceiling $3–8k). Lumpy deal flow, not MRR — the portfolio's only physical-trade, fully uncorrelated play.

First validation test: One farm lot cupped ≥86 pts via a Medellín exporter partner; Maheva emails 25 French independent roasters a FR lot sheet with landed price + EUDR dossier promise. PASS = ≥2 paid sample orders (€75+) AND ≥1 LOI for ≥300 kg. Fail = <5 replies from 25.

#1 kill risk: One quality/logistics failure (moisture, delay, EUDR doc gap) destroys a lot's margin and the roaster relationship in a single shipment.

9. Medellín mid-term apartment — DEMOTED, conditional capital slot

What: COP 470M cash-only amenity unit (gym headline) rented mid-term (30+ days) to nomads. Nightly income is underwritten at $0 without a reglamento expressly authorising tourist use + zoning + RNT — enforcement is real (408 inspections / 105 suspensions since 2024 per W2A; new POT vote Nov 4).

Who: Capital + local manager · Carlos 1–2 hrs/wk.

Cost: $2,000 diligence reserve only. Purchase capital in the $100k envelope: $0 — COP 470M ≈ $148k [calc at Oct 25 FX ~3,171] is a separate decision, taken at the Jan gate, never pre-committed. First money: 6–12 weeks after a close [est].

Monthly potential [est]: Headline ~5.1% net at 82–85% occupancy does not govern. Stress case governs: 65% occupancy / USD-COP 3,600 → ~3.5–4% net ≈ COP 1.37–1.57M/mo, ~$380–435/mo in USD. The 620–800M tiers are lifestyle buys. Local mortgages at 11–14.5% EA are negative carry — cash only.

First validation test: No offer before the Nov 4 POT vote. Then: top-3 buildings' reglamento + uso del suelo + 3-yr admin-fee history in writing, plus mystery-shop 5 mid-term operators. PASS = written tourist authorisation OR mid-term math still ≥4% net under the 65% / 3,600 stress, at ≤COP 470M.

#1 kill risk: Paying a short-term-rental premium for legally-zero nightly income; admin fees (COP 563–965k/mo) are the quiet yield trap.

10. Expat Auto-Entrepreneur Desk — Maheva-owned

What: French-language remote admin desk for French nationals living abroad who keep micro-entrepreneur status: PA e-invoice reception setup (mandate live since Sep 1, 2026, with no walk-in alternative abroad), quarterly URSSAF declarations, cross-border e-reporting. €299 setup + €49–79/mo. Admin/setup help only — not tax advice; that boundary is in the offer.

Who: Maheva · Carlos 1 hr/wk · Maheva 5–8 hrs/wk.

Cost: Test €0–100 / launch €200–400. First money: 1–2 weeks [est].

Monthly potential [est]: €500–1,000 low / €1,500–2,500 base at 6 mo → €2,000–4,500 base at 12 mo (€6k+ only on domestic expansion, which banks own — not planned).

First validation test: Free FR checker in 3 expat groups (Québec / Colombie / Mexique) + 30 DMs, founding offer €99 setup, 10 spots. PASS = ≥3 paid €99 founding setups in 7 days. Fail = <10 checker completions or 0 paid after 30 DMs → kill; do not chase the domestic market.

#1 kill risk: Thin market — expats who kept the status abroad may be hundreds per country, not thousands. Portfolio note: with Ficha, GEO, Cabinet and Primera, this makes five ES/FR language-service offers on the list — the W6 cap binds concurrency (max ~4 active at once), which the sequencing below enforces.

11. RutinaGrid — build-once app, PRIMARY first build (app sprint #1)

What: ES/FR-first local-first habit tracker with a free widget + generous free tier — the wedge is HabitKit's #1 complaint cluster (paywalled widget, 4-habit cap). Category comp: HabitKit $602k in 2025 / ~$28k MRR — claimed, founder-disclosed, not Stripe-verified. Lifetime fenced at ≥3.5× the $14.99 annual hero, never the hero itself.

Who: Carlos + agents build; the AI maintenance loop maintains · Carlos ~20–30 min/wk steady.

Cost: ~$200–700 build cash [est], 4–5 weeks, Android-first. ⚑ TN-flagged: developer-account identity/entity is a real structural choice (whose name store accounts, payouts and tax forms sit in). Flagged per Carlos’s rule, not a precondition imposed by this list. First money: Launch late Nov for the Jan-1 resolution window — feasible only from an Oct-26 / Nov-2 start with a 20-tester pool pre-recruited. A Nov-9+ start loses the window.

Monthly potential [est]: Base (2,500 installs/mo, 2.1% RevenueCat-median conversion) ≈ $937 net/mo at M6 → ~$1,164 at M24; bear ≈ $214/mo. Break-even ≈ 313 installs/mo.

First validation test: The start-date gate itself + closed-test recruitment: PASS = build started by Nov 2, 20 testers recruited in week 1, Android production approved before the Jan window. ASO installs measured in the first 90 days against the 313/mo break-even pace.

#1 kill risk: ASO install volume — revenue is near-linear in installs and conversion is already at the verified median. Plus the streak-loss P0 (repairs shipped in v1) and Samsung/Xiaomi OEM battery-killer widget fragility, which has no dev-side fix.

RutinaGrid app icon ✅ Shipped — v1 feature-complete (Oct 26, 2026). 230 tests green · ES/FR/EN, streak freezes, reminders, display widget · Play closed test starts Dec 8 (production-access application follows ~Dec 24). Screenshots of the real shipped build (Verde theme, seeded demo habits):

RutinaGrid home screen — light theme — Home — light

RutinaGrid home screen — dark theme — Home — dark

12. Cartas FR↔ES — build-once app, first to TEST (app sprint #2)

What: Spaced-repetition vocab app with professionally curated decks for French-for-Spanish-speakers (v1 direction only), offline, no account, no deck export ever. Lifetime $29 + $5–9 packs, never subscription-first. Price evidence: AnkiMobile $24.99 one-time (verified store price); comparative Romance decks €24.99/level (claimed listings). Maheva's native-FR QA is the moat.

Who: Maheva QA (25–40 hrs one-time, the critical path; + a paid second native-FR auditor on a 10% sample) · Carlos ~20–30 min/wk.

Cost: Pre-sale ≤$200 to run / build $250–750 cash, 6–8 weeks elapsed. First money: The pre-sale itself, weeks 2–3 — in parallel with RutinaGrid's build.

Monthly potential [est, conditional on pre-sale PASS]: Base ≈ $2,070 net/mo at M6 → ~$3,730 at M24. Cash break-even is 25–51 buyers ≈ the PASS threshold itself — the only finalist whose gate, if passed, has already paid for its build.

First validation test (the cheapest falsification in either sprint): PASS = ≥50 paid founding orders OR ≥$1,250 gross in 14 days on ≥1,200 qualified visits. KILL = <20 orders AND <$500. Grey zone 20–49 = one 7-day extension only. Deposits don't count. Cap note: this is a product pre-sale under the same Week-0 counsel gate; if counsel counts it as a third sales test, it slips to after the Nov 13 gate — it does not stack on Cabinet + Ficha by stealth.

#1 kill risk: Content quality / Maheva single-point-of-failure — one wrong gender/register teaches an error and kills the native-QA claim. (Duolingo teaches FR to ES speakers free; distribution is SRS communities, francisation and mixed-couple channels.)

Cartas app icon ✅ App complete — FSRS-6 engine + full app built against fixture decks; 88 tests green · content track (Maheva QA) is the remaining path to launch. Screenshots of the real build (tinta y papel theme, fixture deck):

Cartas review card — actuel — Review card — «actuel»

Also cleared the bar (5) — real, sequenced behind the core

13. Operator, Not Coder — ⚑ TN-flagged (restored; W6 froze it on Gate-0 grounds only)

What: Done-for-you AI chief-of-staff install — productising Carlos’s own Mavi-style agent workflow for ES/FR executives. Anchors: CloseBot ~$3k/install; First Movers $15–40k [vendor claims, W5A].
Who runs it: Carlos (design + install) 4–6 hrs/wk · Cost: ~$0–500 · First money: 1–3 weeks [est]; test = 20 network sends, PASS = ≥1 founding deposit + ≥5 calls.
Monthly potential [est]: $5,000–10,000/mo first plateau.
#1 kill risk: ⚑ TN-flagged, the strongest flag on the list — the product is literally his day-job workflow (visa + employer IP). Ranked on merit per Carlos’s rule; the flag is the whole note.

14. DocuES — build-once app, Alternate 2 (app sprint #3)

What: Spanish-first scanner + PDF tools suite (merge/split/compress/sign/OCR), offline, no account, no watermark, no weekly-trial trap. Android-only v1. Mexico reality check: the whole MX scanner pool is only ~$10–17k/week split among incumbents (Sensor Tower estimates, W3).

Who: Carlos + agents, AI loop · Carlos 20–30 min/wk steady (45–60 in month 1).

Cost: $350–1,050 build [est], 5–6 weeks Android-only; iOS only on Android proof. First money: ~6–8 weeks.

Monthly potential [est]: Base (3,000 installs/mo, estimated 1.8% conversion — the softest input in either sprint) ≈ $2,271 net/mo; break-even ≈ 153 installs/mo. ~95% of revenue is lifetime-led new-sale flow: if installs stop, revenue stops within a quarter. The bull case (~$7,362/mo) would be 12–20% of the entire MX scanner pool — struck as a plan, ceiling only.

First validation test: Build only after RutinaGrid's loop has proven days behind it. MVP to closed track on long-tail/trámite ASO terms: PASS = conversion ≥1.5% at a ≥153-installs/mo pace; FAIL = <1% conversion → no iOS build, wind down.

#1 kill risk: A corrupt/lost PDF of an irreplaceable document (worst P0 in either sprint) on a beta, GMS-delivered black-box capture core. Mandatory repairs: atomic-save/verify + 30-day recovery vault ("never the sole copy"), real basic-capture fallback in v1. Apple 4.3(b) rejection risk is the highest store risk if iOS follows.

15. PruebaSocial — build-once app, gated #4 (app sprint #4)

What: WhatsApp-first testimonial + FAQ widget for one LatAm vertical — dental/aesthetic clinics — at $12/mo per location (annual $115/yr default). Category pay of $15–60/mo is proven at pricing level across Senja/Famewall/EmbedSocial/WiserReview; Senja's revenue figures are all claimed and were not used as forecast inputs.

Who: Carlos 30–45 min/wk + Maheva onboarding/support 3–6 hrs/wk at base scale — the heaviest loop finalist. Build last or not at all, and not until RutinaGrid's loop has 90 proven days.

Cost: $250–450 build, infra $30–80/mo. First money: Gated — after the gates below clear.

Monthly potential [est]: Base ≈ $972 MRR at M12 → ~$1,752 at M24. At the Baremetrics-verified 5.8%/mo churn for $10–25 ARPU, $1k MRR needs 84 active clinics + 4.9 adds/mo forever just to stand still; bear (10% churn) never reaches $1k in 24 months.

First validation test — hard gates, all three: (1) Google Business Profile API approval in hand, or the week-8 re-rank trigger fires; (2) ≥50% of the first 10 beta clinics prepay the $115/yr annual — monthly-only demand falsifies the model; (3) MX VAT registration + CFDI costed before the first MX sale. Gates 1+2 both fail → KILL (the reserve, ImprimeYA, is the promotion candidate — conditional, not a validated finalist).

#1 kill risk: Churn + platform gates — a slow policy/churn bleed, the failure mode a weekly loop notices last. Mandatory v1 repairs: Places reviews cached <30 days only, separate Places/GBP pipelines, no review gating.

16. Southbound Setup Desk — Colombia relocation concierge

What: Fixed-fee, asset-light concierge landing one person/family in Medellín correctly in 30 days — visa route via partner lawyers, cédula + bank account, health cover, lease-risk checklist, and a residency-day calendar. $1,495 Setup / $2,995 Premium / $149/mo Guardian. Incumbent proof of willingness to pay: GoNomad charges $3,150–8,000 for housing-bundled equivalents (W5A); nobody in that stack sells the residency-day guardian.

Who: Maheva ops (FR/ES book) + Mom (Canada intake) · Carlos 3–5 hrs/wk — the worst Carlos leverage of any also-cleared entry, which is why it is second-wave despite clearing.

Cost: Test $100–300 / launch ~$800. First money: 2–3 weeks [est].

Monthly potential [est]: $1,500–3,000 low / $3,000–6,000 base at 6 mo → $3,000 / $5,000–8,000 at 12 mo, seasonal (Jan/Sep peaks). Discounted from W5A's $5–12k — trust + regulated-advice routing is unproven.

First validation test: Landing page + posts in 3 Medellín expat/nomad communities + 10 arrival interviews. PASS = ≥2 founding deposits ($250) in 10 days. Fail = people want free advice / only want housing.

#1 kill risk: Trust + regulated advice — clients hear "legal/tax advice" while buying coordination. All visa/tax decisions route to named partner professionals in the contract. Gate 0: medium — entity/invoicing route cleared before first charging.

17. Arrière-Boutique Trilingue — back-office pods, cleared as a paid test

What: Productised asynchronous Spanish back-office pods for Québec/Canadian firms selling to Hispanics — e-commerce returns, ES document intake/chasing, dunning, review replies — sold in French by Maheva, QA'd by her, invoiced in CAD via Mom's CCPC, at CAD 1,200–1,800/mo per pod. Async by design: it survives bad weeks, unlike a phone front desk.

Who: Maheva (sales + QA — the bottleneck) · Carlos 2 hrs/wk · 1–2 Colombian bilingual agents per pod.

Cost: $0–200 outreach / CAD 1,000–2,000 launch. First money: 3–6 weeks [est].

Monthly potential [est]: CAD 0–1,200 low / CAD 3,600–5,400 base at 6 mo (3 pods) → CAD 3,600 / CAD 6,000–9,000 at 12 mo (4–5 pods; W5B ceiling CAD 5–12k).

First validation test — the whole entry is this test: Maheva messages 20 Québec e-commerce/services owners in French: a 2-week Spanish-queue pilot, CAD 600. PASS = ≥1 paid pilot by day 10. Fail = >50% reply "we have no Hispanic customers" → retest Ontario/US Hispanic-facing SMBs once before a full kill. No hiring before it passes.

#1 kill risk: The buyer may not exist at pod scale — a small Québec firm's Spanish queue may be 5 tickets/week, not a pod.

18. Borderline tax copilot — DEMOTE + REPAIR, deliberately last

What: No SaaS, no raise track until proof. ONE fixed-fee productised packet review sold through ONE partner firm on revenue-share. The portfolio's only remaining long/fundraise option, ranked below every cash engine (W6). The $500k–1.5M pre-seed path [est] requires 10 firms / 100 paid packets at ≥90% acceptance and is in no base case.

Who: Carlos (trust sales + review) · 3–5 hrs/wk.

Cost: Validation $500–1,500 [est] (counsel sign-off = milestone zero, plus proper anonymisation of the demo packet) / launch $3,000–8,000 post-pass only. First money: 6–12 weeks realistic [est] — and selling to firms heading into Jan–Apr filing season is the worst quarter to sell.

Monthly potential [est]: Low $0 (the modal outcome: polite calls, zero committed packets) / base $1,000–3,000 lumpy consultancy at 6 mo → $3,000–5,000 if one partner firm feeds steady cases. Confidence flag: no validated packet price exists anywhere in W1–W6 — this is the weakest number in the list.

First validation test: Redacted sample packet + 1-pager to 30 practices (post-tax-season window). PASS = ≥1 firm commits a REAL client packet. Hard kill (standing rule, reaffirmed W6): zero firms commit in the 30-day test = the raise track is dropped permanently.

#1 kill risk: Gate 0, highest in the list (1/5) — his own case is the demo data; no demo use and no outreach naming it before counsel sign-off. TAM + liability is the investor fight behind it.

Conditional watch — NOT counted in the 17

Ontario AODA filing sprint (Mom) — cleared W7 on deadline + Mom leverage (packs CAD 1,500–3,500; Nov–Dec spike $6–15k/mo [est], then a cliff to a monitoring-retainer tail), but it is Mom's third business here. It runs only as a December swap if Primera's founding-5 test has resolved and Mom has capacity; test = ≥3 calls or ≥1 paid deposit from 30 scan-led emails. If that swap doesn't open, it is the first cut — willingness to pay is shaky (the filing is a self-declaration many firms do free in 20 minutes).
ChatGPT Ads early-mover desk (ES/FR) — clears conditional on ONE binary gate checked first: an eligible billing entity (Colombia eligibility UNVERIFIED in W5C; Mom's Canada entity is the candidate route) AND ≥2 brands paying a $500 test fee. No eligible route → it drops off entirely. Modelled $2–4k/mo at 6 mo [est] if the gate passes, discounted hard from W5C.
LangLang ($84.5k ask) — $0 re-login / Play Console verification watch only. Buy solely at verified trailing-6-month net with API-36 + developer-verification compliance shown and a 20–30% holdback. At ask, cut (AccessibilityService / Android 17 restriction risk; 96% margin is a poster claim).
Vertical AI workflows, fintech/identity — watchlist option only: no wedge, no named queue, no design partner, and Persona ($2B) bundles the category. The YC Nov 2 deadline is struck from all planning. Revive only with one named review queue + one paid/LOI design partner; earliest accelerator S27.
VeriFactu migration sprint (Spain) — only the narrowed $0 test survives: a gestoría pays for a pilot AND names ≥10 SLs not already on a VeriFactu-included platform (Holded bundles it free; the ~8% adoption figure is stale Nov-2025 data; the deadline already slipped once via RDL 15/2025). Instant kill on any further postponement.
Operator, Not Coder — IN THE LIST at #13 (restored Oct 25 per Carlos: ⚑ TN-flagged, not frozen).
Lab Second-Read — paid interpretation is CUT (medical practice risk). A navigator-only form (a partner clinic's MD is clinician/seller of record; zero interpretive claims) may be re-proposed only if an MD partner accepts those terms in writing.
AI Longevity Lab (clinic) — not a standalone entry; skincare (#6) is what validates toward it. Phase 0 gates unchanged.
Marketplace watch — screenshot-to-code revives only at ≤12 months of verified profit with churn <8% (at the $32k ask it is cut — the feature is free in v0/Lovable/Bolt). Shopify trust-apps portfolio ($48.9k) + DotMarket FR invoicing SaaS were never red-teamed and remain UNVERIFIED/STALE.

Parallel-portfolio recommendation

The centre of gravity, after the red team: Mom, Maheva and capital run most of it; counsel clears it first; Carlos's ~10 hrs/wk of sales/delivery go only to the tests actually running. The ES/FR services are one correlated bet wearing different costumes (same operators, same Gate 0 entity question, same SMB churn) — treat them as one position sized by operator hours. Genuinely uncorrelated exposures: T-bill interest, the service position, a Colombian product margin (skincare), a COP real-asset option (apartment, uncommitted), physical trade (coffee), and the app portfolio's ASO/lifetime-sale economics.

Sequencing

Week 0 (Oct 26–30) — counsel and verification only, NO paid tests. One counsel engagement scoped to the tests (TN + PIIA + entity/invoicing route per test, and the app developer-account identity/entity — both sprints share this gate). Cabinet clock: counsel engaged with a written sign-off date ≤Nov 6, second FR counsel escalated Oct 29 if no path emerges. In parallel, all ~$0: the accountant withholding answer (sleeve), Mom's designation/EFILE/insurance verification (Primera), Tecnolabs + San Roque RFQs in writing (skincare), the 30-min Acquire/Flippa re-login check (marketplace watch), Ficha's 30-target defect list and Cabinet's ≥20-cabinet warm list (preparation, not tests). Gate G0 at week's end: each test is Cleared / Cleared-with-limits / Blocked — only Cleared tests run.
Weeks 1–2 — exactly two sales tests, per the W6 cap. (1) Cabinet (Maheva ~8 hrs/wk — the only hard external window, Nov 20) and (2) Ficha (Maheva ~5 hrs/wk, Carlos ~2 — fastest cash). Maheva is at ~13 of her 15–20 hrs: no third Maheva test, no skincare clinic visits, no GEO FR outreach. The income sleeve executes in this window once the withholding answer lands (execution, not a sales test). App build track in parallel: RutinaGrid build starts (Oct-26/Nov-2 window, agent-led; Carlos approvals only) and Cartas's pre-sale runs weeks 2–3 under the counsel gate noted in #12. Decision gate G1, Nov 13: Cabinet final bar and Ficha 7-day bar judged; every kill frees exactly one slot.
Months 2–3 (Dec 2026 – Jan 2027) — rotation in fixed queue order. GEO fix sprints first into a freed slot (white-label partner push first — it is GEO's only path to a $5k week). Primera's test enters when a Weeks 1–2 test resolves (Mom-operated, but it counts toward the caps). Skincare's LOI test third, only with supplier answers in writing and Maheva freed from the bilan peak. DocuES builds after RutinaGrid's loop proves itself; PruebaSocial stays gated last. AODA may run in December only as a swap for Primera's slot, never stacked. Property decision window opens after the Nov 4 POT outcome + written proofs + stress pass.
Months 4–12 — scale only what passed. Cabinet buyers convert to subscriptions (setup-credit hook expires Jan 31); the Apr–Jun déclaration season is sold to the same cabinets first. Ficha harvests setups + 90-day blocks; the chain bundle is tested only after single-location economics are proven. GEO's LatAm/EU raise conversation ($250–750k [est]) unlocks only at 50 paying brands / ≥$15k MRR / ≥40% recurring. Primera hires the part-time bookkeeper before client 6. Skincare's reorder, second SKU, the single magnesium-glycinate registro option and the ≤50-unit red-light demo batch all sit behind the ≥60%-sell-through-in-60-days gate. Borderline's firm-channel test runs in its proper post-tax-season window, if it survives at all. App portfolio capped at 3 live apps until the loop has run 90 days on app #1. Standing rhythm: one ≤1-hr gate review per month — active tests ≤2, active language services ≤4, Maheva ≤20 hrs/wk, Carlos sales/delivery ≤10 hrs/wk; any breach pauses the newest entrant, never expands the caps.

Capital split — ~$100k envelope [est plan; ceilings with gates, not spend commitments]

Tranche	Amount	Gate / release condition
Structured income sleeve (control)	$50,000	Accountant withholding answer in hand; ~55% bills/SGOV, ~25% JEPI, ~20% JEPQ inside the tranche; judged on total return vs SGOV-only
Counsel + validation fund	$8,000	Counsel releases Week 0 (it is Week 0); test spend (~$365) releases per test, post-G0 only. App-sprint launch cash for the first 3 apps (~$1,074–3,124 [est] + ~$600 shared Claude) is drawn from this fund and the reserve
Skincare gated inventory (1 SKU)	$12,000	$0 until supplier terms are in writing AND the LOI bar (≥300 units, margin + payment days written) is met
Primera Cuenta / CCPC setup	$3,000	Mom's designation + EFILE + insurance verified in Week 0
Property diligence reserve	$2,000	Diligence only, from Nov 4. Purchase capital: $0 — a passing unit is a separate ~$148k [calc] cash decision
Marketplace buys	$0	Both headliners demoted at ask; the re-login check costs $0. Any revival competes for the reserve against proven winners
Reserve / follow-on for winners	$25,000	Releases only at G1/G2 to a slot that PASSED its bar
Total	$100,000	$75,000 (sleeve + reserve) stays liquid throughout; maximum capital actually at risk before G1 is the counsel/validation tranche (~$8k) plus sleeve market movement

Explicitly $0 in this split, by W6 verdict: red-light container inventory, supplements production, Vertical AI build before a design partner, Borderline SaaS build, Operator delivery before written counsel, any property above COP 470M or on credit.

Considered and cut

Recovery Desk as a standalone — cut (W6): a $199 one-campaign pilot imports the clinic's WhatsApp/consent/list-quality risk for a one-shot product; survives only as a module inside Ficha clinic packages.
Québec front-desk + human-QA hybrid as a standalone — cut (W6): Zenicall holds FR/Loi 25 at CAD 200/mo flat and AI reception starts at $29–49/mo.
Garde de Nuit (per-job pricing) — cut (W1A, confirmed W6): $39/job ≈ $585/mo at 15 jobs = 2–3× the occupied market.
Real-estate follow-up desk, standalone — cut (W1A): AI ISAs at $69/mo / ~$1 per lead killed the human desk in English; ES/FR WhatsApp module of the Outreach Engine only.
Red-light device private label, standalone — cut (W6): returns + warranty on sea-freighted electronics eat the 20–35% contribution; survives only as a ≤50-unit demo batch inside the skincare/Lab brand, after written INVIMA classification + exact-model 510(k) verification.
Supplements own line as an entry business — cut (W6): INVIMA Alert 331-2026 (Oct 20, 2026) + registro backlog make 2–6 months optimistic (budget 6–10); one magnesium-glycinate registro filing survives as a cheap option, only if skincare passes.
Lab Second-Read paid interpretation — cut (W6): individualized lab interpretation is medical practice with a disclaimer stapled to it; navigator-only form is on the watch list.
Screenshot-to-code at $32k — cut at ask (W6): the feature is now free inside v0/Lovable/Bolt and the listing is unverified/stale.
Aquarium ($60k) / wedding platform ($87k) / lease abstraction ($83.8k) / Alpa Cards ($76k) — cut from the finalist pool (W2B + Flippa H1 2026: content −39%): unverified, and each fails a proof gate.
ES AI companion — cut (W3, W6 context): fundable category, worst team fit — Character.AI bankruptcy Oct 2026 + CA SB 243 / NY / EU AI Act companion rules, 24/7 safety load, no distribution edge. The sprint's trap candidate.
Pay-per-call affiliate lead-gen — cut for Carlos (W4A): real payouts (median $32.50/call, benchmark/poster) but paid-traffic arbitrage at 5+ hrs/wk with no US local-traffic asset; monitor only.
Generic digital products / template & prompt packs as a business — cut (W4A): Gumroad median $72/mo across 146k products; packs survive only as add-ons to the Ficha/Cabinet/GEO lists.
YouTube/newsletter as a quick-cash slot — cut on timeline (W4A): sponsorship is priced on views Carlos does not yet have; it stays a phase-2 channel asset.
Medellín offline cash businesses / vending / ATM / direct CO lending / domains & leasing — cut as capital-only (W2C, confirmed W6): all are operator jobs with unverifiable cash, not passive capital.
ACA Spanish lead desk — cut (W5/W6): 76-day seasonal window, highest Gate 0 flag in W5, licensed-partner + TCPA/health-data liability larger than lead revenue.
Fully agent-run / autonomous businesses as a model — cut (W1B): fresh Oct 2026 logbooks show $0 revenue / 0 visitors; agents multiply a human-owned channel, they do not create one.
App-sprint cuts (from its own final list) — EspacioLibre (kill-test failed 0/5), FactureSolo + successor (free PA-registered natives own the job at €0), DevCaja (free DevToys/CyberChef), Suscripciones ($1.99 one-time price ceiling), Días Claros (lifestyle reserve only), ImprimeYA (reserve only — promotes if a finalist fails a hard gate, after its own validation, shipping against ≤5 frozen printer models).

Evidence base: this sprint's wave1.md–wave7.md (hidden_files) and the parallel app sprint's wave1.md–wave6.md + app-shortlist.md. Load-bearing claims keep their wave labels throughout: TrustMRR/Stripe-read GEO leaders, Zenicall and Jobber pricing, Holded's free VeriFactu bundling, INVIMA Alerts 73-2026 / 331-2026, USCIS Jun–Jul 2025 TN guidance (via JD Supra / Stone Oak summaries in W6C), IRCC diaspora figures, Banco de la República remittances, Eurostat coffee imports, RevenueCat conversion medians, Baremetrics churn cohorts, Sensor Tower scanner-pool estimates, HabitKit founder-disclosed revenue (claimed). Every forward number is an [est] planning figure unless a source is named in-line.